<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by NetworkTigers on Medium]]></title>
        <description><![CDATA[Stories by NetworkTigers on Medium]]></description>
        <link>https://medium.com/@networktigers?source=rss-47f1d5f569dc------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*hckdiI-cANdthjPQ8dbGUw.jpeg</url>
            <title>Stories by NetworkTigers on Medium</title>
            <link>https://medium.com/@networktigers?source=rss-47f1d5f569dc------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 23:46:11 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@networktigers/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How do you recover from a ransomware attack?]]></title>
            <link>https://medium.com/@networktigers/how-do-you-recover-from-a-ransomware-attack-a3f7941f021c?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/a3f7941f021c</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[ransomware-attack]]></category>
            <category><![CDATA[network-security]]></category>
            <category><![CDATA[cybercrime]]></category>
            <category><![CDATA[ransomware]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Thu, 19 Nov 2026 05:54:17 GMT</pubDate>
            <atom:updated>2026-11-19T05:54:17.121Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*3WZU-O_gha-Lknml" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><strong>Editor’s Note:</strong> This article was originally published in on <a href="https://news.networktigers.com/network-knowhow/how-do-you-recover-from-a-ransomware-attack/">NetworkTigers News</a> on November 2021 and has been substantially updated to reflect current ransomware recovery guidance, including securing administrative access, validating backups and restoring services without reintroducing attacker access.</p><p><em>A successful restore can return compromised systems to service. Backup selection, identity recovery and reconnection decisions determine what comes back with them.</em></p><p>Recovering from a ransomware attack requires removing the attacker’s access before restored services become reachable again. Reinstalling servers and recovering files will not resolve stolen credentials, unauthorized accounts, or compromised management systems that could let an intruder control the rebuilt environment.</p><p>The recovery plan must establish which infrastructure is trustworthy, which backups are usable, and what evidence is required before each service returns to production. Containment, investigation and recovery planning proceed together, with investigation findings controlling what can safely reconnect.</p><h3>Contain the attack and preserve evidence</h3><p>Isolate affected systems immediately. Where multiple systems or subnets are involved, coordinated network isolation may be necessary. Include wireless connections and remote-access paths. Use a separate, trusted communication channel to coordinate the response if corporate email or collaboration accounts may be compromised.</p><p>Network isolation limits further spread but does not stop encryption already running on a host. Responders must also stop identified ransomware processes.</p><p>Avoid indiscriminate shutdowns. Disconnecting a host preserves volatile evidence that powering it off destroys. The joint federal StopRansomware guide recommends powering down affected devices when network disconnection is not possible and shutdown is needed to prevent further spread.</p><p>Have responders preserve relevant logs, ransom notes, encrypted samples and selected system images and memory captures before rebuilding. Record containment actions and their timing so investigators can distinguish defensive changes from attacker activity.</p><p>Activate the incident response team and involve business owners, legal counsel and the insurer where applicable. In the United States, report ransomware to the FBI through its Internet Crime Complaint Center or a local field office. Do not wait until recovery is complete.</p><h3>Scope the intrusion beyond encrypted machines</h3><p>The encryption event does not establish when the intrusion began or how far it reached. Review identity, VPN, endpoint, firewall and cloud audit records to identify the entry point, compromised accounts and systems accessed before encryption.</p><p>Include domain controllers, virtualization management, backup consoles and remote administration systems in the assessment. A server without encrypted files may still be compromised. Endpoint detection and response provides host telemetry, but it does not replace investigation of identity services, network appliances or cloud administration.</p><p>Close the identified entry route and investigate persistence, including unauthorized accounts and unexpected remote-management tools. Where evidence is incomplete, document the uncertainty and restrict the affected environment accordingly. An absence of alerts is particularly weak evidence when logging was disabled or records have expired.</p><h3>Recover trusted administration</h3><p>Perform recovery administration from devices built from trusted sources. Reusing a compromised administrator workstation can expose replacement credentials as soon as they are entered.</p><p>During containment, disable compromised accounts and revoke supported sessions and tokens. Coordinate subsequent credential rotation with removal of attacker persistence. Changing passwords while the attacker still controls the identity infrastructure can leave the replacement credentials exposed.</p><p>Review more than user passwords. Depending on the intrusion, recovery may require rotating service-account secrets, application credentials and certificates; removing unauthorized application permissions; and correcting attacker-added authentication methods. Verify privileged roles and group membership before restoring access.</p><p>If Active Directory or a federated identity system was compromised, use a platform-specific compromise recovery plan. Directory recovery and signing-key changes affect dependent services and require deliberate sequencing. A general password reset cannot establish that the identity system is trustworthy.</p><h3>Select backups against the intrusion timeline</h3><p>A backup taken before encryption may still contain attacker tools or altered configuration. Compare candidate recovery points with the earliest evidence of compromise, recognizing that the first recorded malicious event may not be the actual entry time.</p><p>Restore candidates into an isolated validation environment. Investigate their contents, scan for malware and test application behavior before approving them. A successful backup job establishes neither that the captured system was clean nor that the application can recover.</p><p>Immutable storage protects retained recovery points against modification or deletion under its configured controls. It does not remove malicious content captured before the backup was written. Separately assess the trustworthiness of backup administration and the availability of required recovery keys.</p><p>If the latest defensible recovery point is older than the business can comfortably accept, make the resulting data-loss decision explicit. Identify records that can be reconstructed from independent sources, and plan how transactions created during temporary operations will be reconciled.</p><h3>Rebuild behind an enforced boundary</h3><p>Rebuild compromised systems from trusted installation media or validated images, and apply relevant security fixes before granting production access. Restoring an entire machine image also restores its configuration and software, including anything malicious present when it was captured.</p><p>Keep the recovery environment separated from the compromised estate. A dedicated VLAN is insufficient if routing, shared administration or remote-support systems preserve access across the boundary. Test which systems can reach recovery hosts and through which management paths.</p><p>Allow only the connections required for recovery and validation. Record the purpose, owner and review point for temporary exceptions. <a href="https://news.networktigers.com/network-knowhow/why-network-segmentation-projects-fail-in-real-environments/">Undocumented application dependencies</a> create pressure to broaden access when services fail to start; investigate the missing dependency before opening communication between entire segments.</p><h3>Restore services in dependency order</h3><p>Prioritize business functions and map the services each requires. A restored application may still be unusable without trusted authentication, DNS, databases, certificates or external integrations. Recovery order should reflect those dependencies and the consequences of continued downtime.</p><p>Define acceptance criteria before reconnecting a service:</p><ul><li>Responders have addressed identified access and persistence mechanisms affecting the service.</li><li>The selected recovery data and application configuration have passed validation.</li><li>Responders have reviewed and remediated required credentials and permissions.</li><li>Logging and security monitoring are operating and visible to responders.</li><li>The service owner has tested essential workflows and has a defined procedure to isolate the service again if suspicious activity appears.</li></ul><p>Reconnect in controlled stages and monitor for renewed suspicious authentication, administrative changes and unexpected connections. Keep unresolved systems quarantined. Partners may also require evidence of remediation before restoring their connections.</p><h3>Separate decryption from incident recovery</h3><p>Free decryptors exist for some ransomware families and versions. Check the No More Ransom project and consult responders or law enforcement about applicability. Preserve the original encrypted files and test an appropriate tool on copies before using it more widely.</p><p>The FBI does not support paying a ransom, and payment does not guarantee data recovery. Even a working decryptor leaves the investigation, removal of attacker access and rebuilding work to be completed.</p><p>Investigate data theft alongside service recovery. Restoring a database does not retrieve copies an attacker has already taken. Legal and communications teams must assess notification obligations and extortion exposure independently of whether operations have resumed.</p><h3>Make the next recovery test harder</h3><p>Use the incident to test whether recovery can succeed when production identity, administrative workstations and the primary backup console are unavailable. Measure the time needed to restore a usable business service, including validation and dependency recovery. That exercise exposes weaknesses a file-restore test cannot: inaccessible credentials, missing configuration, untrusted management paths and decisions nobody has authority to make.</p><h3>Sources</h3><ul><li><a href="https://www.ic3.gov/CSA/2023/231019.pdf">CISA, NSA, FBI and MS-ISAC: StopRansomware guide</a></li><li><a href="https://learn.microsoft.com/en-us/security/ransomware/incident-response-playbook-dart-ransomware-approach">Microsoft: Ransomware incident response approach</a></li><li><a href="https://www.microsoft.com/en-us/security/blog/2020/12/21/advice-for-incident-responders-on-recovery-from-systemic-identity-compromises/">Microsoft: Recovery from systemic identity compromises</a></li><li><a href="https://learn.microsoft.com/en-us/defender-xdr/sop-documentation-template">Microsoft: Compromised identity response procedures</a></li><li><a href="https://learn.microsoft.com/en-us/azure/architecture/security/ransomware-resilient-backup-architecture/">Microsoft: Ransomware-resilient backup architecture</a></li><li><a href="https://www.ncsc.gov.uk/collection/disruptive-cyber-attacks/recovering/2-recovery-ongoing-investigations">NCSC: Recovery and ongoing investigations</a></li><li><a href="https://www.fbi.gov/how-we-can-help-you/common-frauds-and-scams/ransomware">FBI: Ransomware reporting and ransom guidance</a></li><li><a href="https://www.nomoreransom.org/en/decryptiontools.html">No More Ransom: Decryption resources</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About NetworkTigers. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a3f7941f021c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The future of hardware security in network infrastructure]]></title>
            <link>https://medium.com/@networktigers/the-future-of-hardware-security-in-network-infrastructure-c6773521577b?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/c6773521577b</guid>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[attestation]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[hardware-security]]></category>
            <category><![CDATA[network-security]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Wed, 18 Nov 2026 05:25:31 GMT</pubDate>
            <atom:updated>2026-11-18T05:25:31.614Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*EbwZaY62LzjSn5GL" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>Hardware security’s next challenge is turning protected keys, platform measurements and component identities into evidence that networks can act on. First published October 27, 2026, on </em><a href="https://news.networktigers.com/hardware-hub/future-of-hardware-security-in-network-infrastructure/"><em>NetworkTigers News</em></a><em>.</em></p><p>Hardware security could play a larger role in deciding which devices join a network, which workloads receive sensitive data and whether infrastructure remains trustworthy after a change. Secure boot and protected key storage already provide foundations. The next step is making hardware-backed evidence useful across distributed systems and throughout equipment lifecycles.</p><p>That development depends on more than adding security features to silicon. Platforms need compatible ways to report their state, management systems need policies that act on the evidence, and cryptographic protections need room to evolve.</p><h3>Attestation in routine access decisions</h3><p>Secure boot can prevent unauthorized components from loading during startup. Measured boot records cryptographic measurements that can be evaluated later. Hardware roots of trust provide a protected foundation for these functions, alongside device identity and key protection.</p><p>These are established capabilities. Cisco’s 2019 white paper on its 8000 Series routers describes hardware-anchored boot verification and a Trust Anchor module with protected storage and device identity. The opportunity ahead lies in connecting such capabilities to operational decisions across infrastructure.</p><p>Remote attestation lets a platform provide cryptographically protected evidence of its state. A verifier evaluates that evidence against policy. Management systems can then use the result before issuing credentials, admitting a device to a workload pool or releasing sensitive data.</p><p>Making platform integrity a routine condition of access requires fresh evidence, maintained reference values and clear responses to failed checks. A legitimate firmware update can change expected measurements, so attestation policies must accommodate authorized change.</p><p>Attestation also has a defined scope. It supports conclusions about the state represented by its evidence; ongoing <a href="https://news.networktigers.com/network-knowhow/think-like-a-network-monitoring-specialist/">network monitoring</a> remains necessary for subsequent activity and conditions outside that coverage.</p><h3>Hardware-backed isolation for infrastructure workloads</h3><p>Confidential computing extends hardware protection to data in use through attested trusted execution environments. Depending on the implementation, these environments can isolate workload memory and execution from the host operating system or hypervisor.</p><p>Hardware-backed isolation can protect sensitive applications on shared servers, including those running network functions or edge services. Before releasing credentials or data, organizations can use attestation to check that the application runs in an environment that meets their security requirements.</p><p>The protection comes from hardware-enforced boundaries and may include encrypted regions of system memory. Its value depends on which components remain trusted, which attacks the implementation addresses and how applications use the protected environment.</p><p>For network teams, wider availability could expand placement options for sensitive services. Adoption will depend on platform support, performance and integration with orchestration and key-management systems. Availability across compute platforms and network appliances will determine where those options become practical.</p><h3>Protected keys for machine identity</h3><p>Device and workload identities depend on cryptographic credentials. Hardware-backed key protection can reduce exposure when software is compromised, but the protection must cover how keys are used as well as where they are stored.</p><p>Device-level trust components can protect local credentials and support platform identity. Dedicated HSMs can provide centralized key protection and cryptographic operations for certificate issuance, signing and other enterprise services.</p><p>A useful direction for future infrastructure is closer integration between identity, attestation and key release. Access to a credential could depend on evidence that the requesting environment meets an approved policy, rather than authentication alone.</p><p>Implementation details remain decisive. Exporting a protected key into an unprotected process changes its exposure. Keeping operations inside a protected boundary reduces extraction risk, while authorization controls must still govern which applications can request those operations.</p><h3>Component assurance across the hardware lifecycle</h3><p>A platform can contain multiple processors, controllers, accelerators and firmware components. As these dependencies grow more complex, assurance must account for the assembled system and the interfaces between its parts.</p><p>NIST’s 2026 workshop report on next-generation secure hardware identifies priorities including cryptographic component identities, lifecycle traceability and interoperable assurance frameworks. These are research and standards priorities, but implementation varies across products.</p><p>Chiplet architectures illustrate the challenge. Combining multiple semiconductor dies in one package raises questions about component authenticity and whether security guarantees hold across interfaces and suppliers. Component-level evidence can help assess those dependencies, provided it can be verified and interpreted in the context of the complete system.</p><p>For infrastructure teams, progress would mean clearer vendor evidence about component provenance, verification coverage and changes during service. Maintaining that assurance also requires authenticated firmware updates, controlled recovery and protection against unauthorized rollback to older vulnerable images.</p><h3>Cryptographic adaptability in platform design</h3><p>The transition to <a href="https://news.networktigers.com/network-knowhow/quantum-resistant-encryption-preparing-networks-for-the-post-quantum-era/">post-quantum cryptography</a> reaches into firmware authentication, device identity, HSMs and embedded cryptographic functions. Long-lived equipment may need to accommodate algorithm changes during its service life.</p><p>Fixed hardware implementations can constrain that transition. A platform may support updated network protocols while retaining a boot-verification mechanism whose algorithms cannot change. Cryptographic adaptability therefore needs to be considered across the platform.</p><p>Future designs may combine updateable implementations with acceleration for demanding operations. Each approach involves trade-offs in performance, flexibility, and protection. Larger keys, signatures or other cryptographic data can also affect memory, storage and protocol behavior.</p><p>For network teams, migration planning will extend beyond protocol settings to the capabilities of embedded components. Update paths, resource limits, and vendor support will determine whether a platform can support new algorithms throughout its expected service life.</p><h3>From hardware features to infrastructure decisions</h3><p>The direction is toward hardware that supplies more usable evidence about its identity, state and protection boundaries. How far that develops will depend as much on interoperability and operational adoption as on chip design. For network teams, the meaningful advance is the ability to use that evidence when placing workloads, granting access and planning the next infrastructure refresh.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/dam/en/us/products/collateral/routers/8000-series-routers/trustworthy-framework.pdf">Cisco: A Trustworthy Framework for Cisco 8000 Series Routers</a></li><li><a href="https://www.rfc-editor.org/rfc/rfc9334">IETF RFC 9334: Remote ATtestation procedureS (RATS) Architecture</a></li><li><a href="https://confidentialcomputing.io/wp-content/uploads/sites/10/2023/03/CCC_outreach_whitepaper_updated_November_2022.pdf">Confidential Computing Consortium: Hardware-Based Trusted Execution for Applications and Data</a></li><li><a href="https://csrc.nist.gov/pubs/ir/8615/final">NIST IR 8615: Workshop on Rolling Next-Generation Secure Hardware into Standards</a></li><li><a href="https://csrc.nist.gov/pubs/sp/800/193/final">NIST SP 800–193: Platform Firmware Resiliency Guidelines</a></li><li><a href="https://csrc.nist.gov/pubs/cswp/39/upd1/considerations-for-achieving-crypto-agility/final">NIST CSWP 39, Update 1: Considerations for Achieving Crypto Agility: Strategies and Practices</a></li></ul><p>*****</p><p>By Ben Walker, a <a href="https://www.networktigers.com/">NetworkTigers</a> Author.</p><p>Ben Walker is a freelance research-based technical writer. He has worked as a content QA analyst for AT&amp;T and Pernod Ricard.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c6773521577b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[7 interface statistics that can point engineers to the wrong fix]]></title>
            <link>https://medium.com/@networktigers/7-interface-statistics-that-can-point-engineers-to-the-wrong-fix-8e675ca6d314?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e675ca6d314</guid>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[network-optimization]]></category>
            <category><![CDATA[juniper]]></category>
            <category><![CDATA[cisco]]></category>
            <category><![CDATA[network-engineering]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 19:48:45 GMT</pubDate>
            <atom:updated>2026-11-16T19:48:45.329Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*fALZNrD-wWZB43Rk" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>A lightly loaded port can drop traffic, while a large error count can describe a problem that ended months ago. The diagnostic value depends on what each measurement actually captures. First published October 23, 2026, on </em><a href="https://news.networktigers.com/network-knowhow/interface-statistics-that-can-point-engineers-to-the-wrong-fix/"><em>NetworkTigers News</em></a><em>.</em></p><p>Interface statistics can be accurate and still support the wrong diagnosis. A traffic average hides the burst that exhausted a queue. A bundle total conceals an overloaded member. An error counter preserves evidence long after the fault has disappeared.</p><p>The problem starts when a measurement is treated as proof of something outside its scope, driving a capacity upgrade, hardware replacement or incident closure that the evidence does not justify.</p><h3>1. Average interface utilization</h3><p>A five-minute utilization figure answers how much traffic crossed the interface over that interval. It cannot show whether the egress queue had enough space throughout.</p><p>Egress-buffer exhaustion can cause a <a href="https://news.networktigers.com/network-knowhow/your-network-switch-can-run-out-of-capacity-with-bandwidth-to-spare/">switch to drop packets despite spare average bandwidth</a>. Drops occur during the burst, while the five-minute average shows headroom.</p><p>Shortening the polling interval improves resolution but does not necessarily expose millisecond-scale congestion. A maximum calculated from one-minute samples is still the maximum of one-minute averages.</p><p>When drops accompany apparently spare capacity, examine hardware queue drops and, where supported, queue occupancy or microburst telemetry. The useful question is whether arrivals exceeded the queue’s ability to absorb and drain traffic during the affected period.</p><h3>2. Aggregate port-channel utilization</h3><p>A four-member 10 Gbps bundle carrying 10 Gbps in one direction appears only 25% utilized against its nominal 40 Gbps capacity. That total could nevertheless represent one saturated member and three idle ones.</p><p>With conventional flow-based hashing, traffic sharing the selected hash inputs follows the same member. The distribution depends on those inputs and the traffic mix; the aggregate percentage does not reveal whether individual members have usable headroom.</p><p>Read the logical interface alongside member utilization, member drops and actual participation. These measurements expose <a href="https://news.networktigers.com/network-knowhow/6-times-a-port-channel-is-the-wrong-solution/">port-channel limitations</a> that the bundle total conceals. Adding links without examining the distribution can increase the advertised capacity while leaving the affected flows constrained.</p><h3>3. A single full-duplex utilization percentage</h3><p>A dashboard labeled “interface utilization” may combine receive and transmit traffic in ways that obscure the limiting direction.</p><p>Consider a full-duplex 10 Gbps port receiving 9 Gbps and transmitting 1 Gbps. Adding both directions and dividing by 20 Gbps produces 50%. Yet the receive direction is already at 90%. Adding them and dividing by 10 Gbps instead yields 100%, even though neither direction is fully occupied.</p><p>Both calculations can be performed correctly while answering an unhelpful capacity question. Cisco’s SNMP utilization guidance recommends measuring input and output separately.</p><p>Keep directional rates visible and establish the denominator. A percentage against physical port speed also says little about headroom beneath a lower configured shaper or contracted service rate. Label those limits separately so a healthy-looking physical interface does not conceal a service bottleneck.</p><h3>4. Zero interface errors</h3><p>Zero errors do not establish zero packet loss. The IF-MIB distinguishes packets discarded without a detected error from packets that could not be received or transmitted because of errors. Buffer pressure is one documented reason for discards.</p><p>A clean error graph can therefore coexist with increasing discard counters. It also describes only the measured interface, not successful delivery across the entire path.</p><p>Preserve the distinction between errors and discards in monitoring. When drops are present, check the platform’s queue and forwarding counters to identify where they occur and why. Replacing an optic because traffic is being discarded can leave the actual congestion untouched.</p><h3>5. Cumulative error totals</h3><p>A port showing 50,000 errors looks worse than one showing 50. Without an observation window, that comparison has little diagnostic value.</p><p>The larger total might have accumulated during a cabling fault repaired months ago. The smaller count might have appeared during the last few seconds. The distinction determines whether the reading describes history or an active fault.</p><p>Counter scope also varies. For example, Juniper’s documented interface-error monitor reports CRC errors and link flaps since monitoring began. A monitoring-session total cannot be compared directly with a counter accumulated over a different period.</p><p>Capture timestamps and changes over a defined interval before clearing anything. Compare the increase with traffic volume over that same interval where a meaningful error ratio is available. Preserve the original evidence so another engineer can reconstruct when the problem occurred.</p><h3>6. Traffic rates derived from 32-bit counters</h3><p>A plausible traffic graph can be wrong because the underlying octet counter wrapped more than once between polls.</p><p>A 32-bit octet counter cycles after approximately 4.29 billion bytes. Cisco documents a minimum wrap time of roughly 34 seconds at 1 Gbps with back-to-back full-size packets. A five-minute polling interval can therefore span several wraps.</p><p>The difference between two readings cannot reveal how many complete cycles occurred. A collector that handles a single wrap can still undercount traffic when several occurred, producing an apparently credible rate.</p><p>Use supported 64-bit high-capacity octet counters for high-speed interfaces and verify that the monitoring system is collecting those counters. SNMPv2c or SNMPv3 is required to retrieve Counter64 objects; selecting a newer protocol alone does not guarantee that the monitoring template uses them.</p><p>Counter resets require separate handling. RFC 2863 provides ifCounterDiscontinuityTime so collectors can detect discontinuities and avoid deriving a normal traffic rate across them.</p><h3>7. Corrected FEC error counts</h3><p>A rising forward error correction count can look like evidence of packet loss when it records errors the receiver successfully corrected.</p><p>Corrected and uncorrected errors must remain separate. Juniper’s interface monitoring documentation exposes both, alongside pre-FEC bit error rate and FEC mode. Combining them into a single “errors” alarm discards the distinction that makes the telemetry useful.</p><p>A corrected-error total alone does not establish application impact. Equally, successful correction is no reason to ignore a changing physical-link condition. Examine the trend alongside uncorrected errors, CRC errors and the applicable link specifications.</p><p>Check the unit before comparing devices: a count of corrected bits and a count of corrected codewords describe different quantities. Thresholds must match the hardware, FEC mode, and counter definition rather than treating every nonzero value as a failed link.</p><h3>Give every alarm a measurement boundary</h3><p>A useful monitoring review starts with the decision each alarm is supposed to support. A capacity alarm needs the relevant direction and limiting rate. A physical-fault alarm needs the appropriate error category and observation window. A congestion investigation needs visibility into the queue or member carrying the affected traffic.</p><p>Put those definitions into dashboard labels and alert descriptions. An engineer responding at 2 a.m. should not have to reverse-engineer what “utilization” or “errors” means before deciding whether to change the network.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-9300-series-switches/225791-frequently-asked-questions-output.html">Cisco: Output drops on Catalyst 9000 switches</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/lan-switching/etherchannel/12023-4.html">Cisco: EtherChannel load balancing and redundancy</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/ip/simple-network-management-protocol-snmp/8141-calculate-bandwidth-snmp.html">Cisco: Calculating bandwidth utilization using SNMP</a></li><li><a href="https://www.rfc-editor.org/rfc/rfc2863.html">RFC 2863: Interface counters, discards and discontinuities</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/ip/simple-network-management-protocol-snmp/26007-faq-snmpcounter.html">Cisco: SNMP counter widths and wrap intervals</a></li><li><a href="https://www.juniper.net/documentation/us/en/software/junos/cli-reference/topics/ref/command/monitor-interface-error.html">Juniper: Interface error monitoring and FEC fields</a></li><li><a href="https://apps.juniper.net/feature-explorer/feature/7875">Juniper: FEC monitoring sensors and counter units</a></li><li><a href="https://www.juniper.net/documentation/us/en/hardware/ptx-module-reference/topics/concept/pic-ptx-series-100-ge-dwdm-otn.html">Juniper: PTX FEC corrected-bit and uncorrectable-codeword definitions</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at NetworkTigers.com.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e675ca6d314" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your network switch can run out of capacity with bandwidth to spare]]></title>
            <link>https://medium.com/@networktigers/your-network-switch-can-run-out-of-capacity-with-bandwidth-to-spare-d1a7e22fbb86?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/d1a7e22fbb86</guid>
            <category><![CDATA[network-traffic]]></category>
            <category><![CDATA[switch]]></category>
            <category><![CDATA[cisco]]></category>
            <category><![CDATA[hardware]]></category>
            <category><![CDATA[network-architecture]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Wed, 11 Nov 2026 20:06:10 GMT</pubDate>
            <atom:updated>2026-11-11T20:06:10.999Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*1ce1qzwDq0EYBEgp" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>A switch can have room for more traffic but no room for another forwarding entry or security policy. Hardware resource limits determine which changes it can actually support. First published October 20, 2026, on </em><a href="https://news.networktigers.com/network-knowhow/your-network-switch-can-run-out-of-capacity-with-bandwidth-to-spare/"><em>NetworkTigers News</em></a><em>.</em></p><p>A segmentation change can interrupt forwarding on a switch whose uplinks are barely busy. A new route can appear in software without being successfully installed in hardware. In both cases, bandwidth monitoring can miss the resource that has reached its limit.</p><p>Switch capacity includes the hardware state needed to forward and classify traffic: MAC addresses, route prefixes, next-hop information and access control entries. Those resources have limits independent of how many bits cross an interface. A network can exhaust them by adding destinations or policies without substantially increasing traffic.</p><h3>The forwarding table has its own workload</h3><p>A throughput specification describes how quickly a switch can move traffic under stated conditions. It does not establish how many distinct forwarding decisions the hardware can represent.</p><p>Nor does every entry occupy one interchangeable slot. Switch ASICs use different memory structures for different lookups. Cisco’s Catalyst 9000 documentation distinguishes hash-based exact-match memory from ternary content-addressable memory, or TCAM. Allocation and use vary by ASIC generation.</p><p>The result is more complicated than a single “table utilization” percentage. Free capacity in one resource does not necessarily help a feature that has exhausted another.</p><p>Cisco documents this explicitly for UADP 3.0-based switches. Their optimized forwarding memory handles exact-match and longest-prefix-match entries, with TCAM serving overflow and exception cases. A combination of route scale, prefix-length variety and resource allocation can exhaust route TCAM while the larger exact-match/longest-prefix-match resource still has space.</p><p>That makes routing table shape relevant to capacity. Two deployments with similar route counts can place different demands on the same platform. A published maximum cannot fully describe a production route mix.</p><h3>A short ACL can have a larger hardware footprint</h3><p>Security policy introduces another translation problem: configuration lines are not a reliable count of hardware entries.</p><p>On Catalyst 9000 switches, Layer 4 port ranges and inequality matches can consume value comparison units. Cisco documents that when these resources or applicable operation limits are exceeded, the software can expand an ACL into additional access control entries that perform equivalent matching. That expansion can then exhaust TCAM.</p><p>The operational consequence can be abrupt. Cisco’s documented ACL exhaustion behavior is that it cannot program a newly added ACL, and the interface to which it is attached drops traffic until resources become available. This also applies when you add entries to an ACL already applied to an interface: if insufficient TCAM resources leave the ACL unloaded, existing entries also fail, and the affected interface drops all traffic, including control packets, until the issue is resolved. This is a specific platform behavior, not a universal rule for every switch.</p><p>For change planning, the relevant measurement is the proposed policy’s hardware footprint on the intended platform. Counting rules in a text file misses the compilation step that determines whether the policy fits.</p><h3>Software visibility does not prove hardware installation</h3><p>Resource exhaustion can create a misleading diagnostic picture because software and hardware views need not agree.</p><p>Arista’s EOS documentation includes a mechanism for identifying learned MAC addresses present in software but missing from hardware on supported Trident platforms. It also documents a MAC-learning failure message for hardware resource exhaustion. A software table entry alone therefore does not establish that the forwarding hardware can use it.</p><p>Routing has a similar verification requirement. Cisco’s IPv4 troubleshooting guidance compares route counts with hardware-installed counts and identifies resource-allocation failures during FIB programming. A healthy routing adjacency does not prove every required prefix reached the ASIC.</p><p>Prefixes are only part of the Layer 3 resource picture. Arista exposes separate usage information for next hops, ECMP groups and ECMP members, among other resources. Remaining route capacity does not establish that all the supporting forwarding objects have room to grow.</p><p>These distinctions help explain selective failures. Existing hardware entries may remain usable while the switch cannot install an additional destination or policy. Troubleshooting needs to connect the affected traffic to the missing resource, rather than treating the switch as uniformly overloaded.</p><h3>Resource profiles move the limit</h3><p>Some switches allow administrators to change how hardware resources are allocated. Cisco’s Switch Database Management templates, for example, assign capacity to supported forwarding and ACL features.</p><p>This flexibility makes scale figures conditional. The maximum for one feature may depend on a profile that leaves less capacity for another. Procurement should establish the supported feature and scale combination, rather than assume every advertised maximum is available simultaneously.</p><p>Reallocation also has an operational cost. Cisco’s Catalyst 9300 IOS XE 17.18 documentation specifies that template changes take effect after a reload. An allocation change that resolves the constraint may therefore require a maintenance window.</p><p>When <a href="https://news.networktigers.com/hardware-hub/what-to-consider-when-buying-a-used-network-switch/">evaluating a switch for purchase</a>, identify the exact model, software release and resource profile behind its scale figures. Port density and uplink speed cannot establish whether it will accommodate the intended routing and policy workload.</p><h3>Buffers reveal a different monitoring blind spot</h3><p>Low average utilization can also accompany packet-buffer exhaustion, but through a different mechanism. Several ingress ports can briefly deliver traffic toward one egress port faster than that port can transmit it. The excess must wait in a finite buffer; a sufficiently large burst causes drops.</p><p>Here, the output is congested during the burst even though a longer measurement interval shows spare bandwidth. Cisco’s microburst analysis explains how ordinary polling smooths away these short events.</p><p>The distinction determines what to investigate. Queue occupancy and output discards help expose burst-related loss. Hardware installation failures and resource counters point toward table exhaustion. Increasing an ACL allocation cannot resolve an overflowing output queue.</p><h3>Budget for the next change</h3><p>Capacity planning should account for the next routing or policy change as explicitly as the next traffic increase. Establish current hardware usage, measure what a representative change adds and confirm that the required entries were installed successfully.</p><p>Leave enough headroom for the changes the network is expected to absorb. A switch that forwards its current workload comfortably can still be unable to accommodate the next deployment. That limit belongs in the upgrade plan before it appears in a failed change window.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-9500-series-switches/217703-understand-hardware-resources-on-catalys.html">Cisco: Hardware resources on Catalyst 9000 switches</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-9300-series-switches/217714-understand-ipv4-hardware-resources-on-ca.html">Cisco: IPv4 hardware resources and forwarding-table limits</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-9500-series-switches/217266-validate-security-acls-on-catalyst-9000.html">Cisco: ACL validation and resource exhaustion</a></li><li><a href="https://www.arista.com/en/um-eos/eos-system-event-logging">Arista: MAC-learning failure detection</a></li><li><a href="https://www.arista.com/en/um-eos/eos-ipv4">Arista: Layer 3 hardware resource monitoring</a></li><li><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9300/software/release/17-18/configuration_guide/sys_mgmt/b_1718_sys_mgmt_9300_cg/configuring_sdm_templates.html">Cisco: Catalyst 9300 SDM templates, IOS XE 17.18</a></li><li><a href="https://blogs.cisco.com/datacenter/nanosecond-buffer-visibility-with-hardware-based-microburst-detection">Cisco: Buffer visibility and microburst detection</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d1a7e22fbb86" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What to look for when buying a used network switch]]></title>
            <link>https://medium.com/@networktigers/what-to-look-for-when-buying-a-used-network-switch-71ed3d71cd0f?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/71ed3d71cd0f</guid>
            <category><![CDATA[network-switches]]></category>
            <category><![CDATA[network-equipment]]></category>
            <category><![CDATA[network-design]]></category>
            <category><![CDATA[business-costs]]></category>
            <category><![CDATA[it-cost-management]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Tue, 10 Nov 2026 11:19:38 GMT</pubDate>
            <atom:updated>2026-11-10T11:19:38.186Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*1u6vZg5tJYynerF2" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><strong>Editor’s Note:</strong> This article was originally published in February 2021 on <a href="https://news.networktigers.com/hardware-hub/what-to-consider-when-buying-a-used-network-switch/">NetworkTigers News</a> and substantially revised in October 2026 to reflect current guidance on buying used network equipment, including configuration, licensing, support and supplier selection.</p><p><em>A used network switch delivers value when its purchase price, required components, software entitlements and remaining service life add up to a lower cost of meeting the deployment’s requirements, leaving budget available for capacity, spares and resilience.</em></p><p>That calculation starts with a complete specification. A switch for an existing stack, a standalone access switch, and a laboratory unit can share a product family but require different components and support arrangements.</p><h3>Confirm what the supplier will deliver</h3><ul><li><strong>Turn the model description into an itemized configuration.</strong> Ask for the full part number and a list of included power supplies, uplink modules, fans and accessories. A product-family name or stock photograph can leave important configuration details unresolved. Missing rack hardware is an inconvenience. A missing uplink module or incompatible power supply can stop the installation. Confirm whether optics, stacking cables and mounting hardware are included, and price any additions before comparing offers.</li><li><strong>Tie availability to the exact order.</strong> “In stock” needs to cover the required configuration and quantity. If five switches are available but only three have the specified uplink module, the order is not ready for a five-switch deployment. Establish whether the supplier holds the equipment or sources it after purchase. Obtain a shipping commitment and agree that substitutions require approval. This prevents an apparently equivalent model from arriving with different stacking, power or software requirements.</li><li><strong>Ask what testing establishes.</strong> A switch can boot successfully while individual ports or PoE functions remain faulty. Ask which interfaces and functions the supplier tests, including uplinks, stacking interfaces, fans and power supplies. “Used” and “refurbished” do not specify a uniform preparation process. Establish what was inspected, tested or replaced. Documentation of those checks is more useful than cosmetic grading alone.</li><li><strong>Assess how the supplier handles discrepancies.</strong> Look for customer experience with accurate descriptions, missing components and replacement requests. A review describing how a reseller resolved a configuration error tells you more about after-sales service than a delivery rating.</li><li><strong>Confirm the equipment’s condition.</strong> A reputable reseller clearly identifies equipment as new, used or refurbished and explains how it assessed its condition. That transparency helps buyers avoid <a href="https://news.networktigers.com/all-articles/fraud-delivers-price-war-as-counterfeiters-think-inside-the-box/">used equipment fraudulently repackaged as new</a>.</li><li><strong>Match warranty arrangements to recovery requirements.</strong> Check the <a href="https://www.networktigers.com/pages/terms-of-sale">warranty coverage and return conditions</a>, including exclusions, shipping responsibilities and whether replacement requires the faulty unit to arrive first. A warranty exchange can resolve a hardware failure without meeting the network’s restoration target. If a site cannot wait for shipping, include an on-site spare in the comparison. The lower acquisition cost of used equipment can make that spare affordable.</li></ul><h3>Check the capabilities of the supplied configuration</h3><ul><li><strong>Verify compatibility at the model and software level.</strong> For an existing stack, confirm supported model combinations, stacking hardware and software requirements. Similar model names do not establish that two switches can operate together. Check required routing, authentication, access-control and management functions against the exact model, software release and license tier. Give the reseller those requirements so its compatibility advice can address the actual deployment.</li><li><strong>Check airflow compatibility.</strong> For equipment with different airflow options, confirm that fans and power supplies match each other and the rack’s cooling arrangement. Correcting an airflow mismatch after delivery adds cost and delays staging.</li><li><strong>Size ports and uplinks for the workload.</strong> Specify endpoint speeds and expected traffic through the uplinks. A 48-port switch can accommodate the endpoint count but may lack the uplink capacity their combined traffic requires. An established model can be the right purchase when its interfaces and forwarding capabilities meet the requirement. Paying for a newer generation has little value if the deployment will not use the additional capability.</li><li><strong>Calculate PoE capacity across the switch.</strong> Check both the PoE standard each endpoint requires and the total power budget with the supplied power configuration. Per-port wattage is not the total available budget. A switch that supports the required power on an individual port may still be unable to power every planned access point, phone or camera simultaneously. Calculate the intended load and allow for planned additions. If the design requires power-supply redundancy, verify that the remaining supply can sustain the required load after a failure.</li></ul><h3>Establish software access and remaining service life</h3><ul><li><strong>Separate installed features from software entitlement.</strong> Ask which software release is installed, which functions are enabled and what licenses or subscriptions the sale includes. Establish how you will obtain the releases needed to maintain the switch. The distinction between hardware ownership and <a href="https://news.networktigers.com/all-articles/the-confusing-world-of-used-network-equipment-licensing-or-how-oems-could-make-money-from-the-secondary-market/">used network equipment licensing</a> affects the total purchase cost. Verify current terms for the specific product rather than assuming an installed feature transfers with the chassis. Cisco’s published policy states that its standard hardware warranty and service contracts do not transfer between owners. For secondary-market equipment to qualify for a new Cisco support contract, it must have valid software licensing and pass a hardware inspection. If manufacturer support is required for deployment, establish eligibility and cost before ordering.</li><li><strong>Confirm cloud registration where applicable.</strong> For cloud-managed equipment, verify that the device can be added to your organization and identify the required subscription. The previous owner must unclaim secondhand Meraki equipment before the buyer can claim it. Arrange to verify registration during acceptance testing so the hardware is ready for configuration.</li><li><strong>Compare lifecycle dates with the deployment period.</strong> End of sale, end of software maintenance and last date of support are different milestones. Check the notices for the exact hardware and relevant software release. A switch no longer sold by its manufacturer can still meet a deployment’s requirements. The useful comparison is between the service period you need and the updates and support available during that period. Equipment for a temporary migration or lab has a different purchasing case from a switch intended for five years of production use.</li></ul><h3>Compare total costs and test before deployment</h3><p>Add required modules, optics, cables, power supplies and software to the quoted hardware price. Include recurring subscriptions and expected electricity costs over the planned service period.</p><p>Account for compatibility with the installed network. Common spares, established configurations and familiar management tools can reduce implementation and maintenance work.</p><p>When the equipment arrives, verify part numbers and accessories, inspect for shipping damage and confirm administrative access. Remove previous configurations and credentials using the vendor’s documented procedures.</p><p>Complete any necessary <a href="https://news.networktigers.com/network-knowhow/tips-for-surviving-firmware-upgrades-on-production-networks/">firmware upgrade checks</a> during staging. Test the interfaces and functions the deployment will use, including PoE, stacking and power-supply failover where relevant. Finish acceptance testing within the return window and before the maintenance window.</p><p>The savings are most useful when they fund something the original budget could not accommodate: a prepared spare at a remote site, a staging environment or additional capacity. Calculate those options alongside the switch purchase, and the secondary market can improve deployment while reducing cost.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/en/us/products/hw-sw-relicensing-program.html">Cisco: Hardware Inspection and Software Relicensing Program</a></li><li><a href="https://www.cisco.com/c/en/us/products/eos-eol-policy.html">Cisco: End-of-Life Policy</a></li><li><a href="https://documentation.meraki.com/Platform_Management/Dashboard_Administration/Operate_and_Maintain/Inventory_and_Devices/Cisco_Meraki_Devices_purchased_Second_Hand">Cisco Meraki: Devices Purchased Second Hand</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=71ed3d71cd0f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[6 PoE myths that can damage hardware]]></title>
            <link>https://medium.com/@networktigers/6-poe-myths-that-can-damage-hardware-77761d2def29?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/77761d2def29</guid>
            <category><![CDATA[network-equipment]]></category>
            <category><![CDATA[network-cabling]]></category>
            <category><![CDATA[power-over-ethernet]]></category>
            <category><![CDATA[connector]]></category>
            <category><![CDATA[hardware]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Tue, 03 Nov 2026 19:15:58 GMT</pubDate>
            <atom:updated>2026-11-03T19:15:58.023Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*GpCRtT5ZtQzWNou_" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>A familiar connector and a PoE (Power over Ethernet) label can conceal critical differences in voltage, compatibility and protection. First published October 13, 2026, on </em><a href="https://news.networktigers.com/network-knowhow/poe-misunderstandings-that-can-damage-hardware/"><em>NetworkTigers News</em></a><em>.</em></p><p>A PoE device gets both data and electrical power through a single Ethernet cable. A device that will not power up is rarely in danger. However, insufficient power typically causes allocation denial, reduced functionality or repeated restarts. Permanent damage comes from the other direction: incompatible voltage, bypassed detection, or electrical load the cabling and connectors were never assessed for.</p><p>Standards-compliant IEEE 802.3 PoE detects a compatible powered device before applying operating power. That protection does not extend automatically to every injector, splitter or port setting carrying a PoE label, and a working link proves nothing about it.</p><h3>Myth 1. All PoE injectors are interchangeable</h3><p>An RJ45 connector and a PoE label do not establish electrical compatibility. IEEE 802.3af, 802.3at and 802.3bt define standardized power delivery, but passive PoE products can use different voltages, pin assignments and polarities.</p><p>A passive injector can apply power without first establishing that the connected equipment can accept it. Replacing a 24-volt passive injector with a 48-volt passive unit can permanently damage a device limited to the lower voltage. Matching the manufacturer alone is insufficient when its product range includes different powering methods.</p><p>The Ethernet Alliance specifically warns that incompatible passive PoE equipment can cause permanent damage from excessive voltage. An IEEE-compliant source connected to a passive-only device lacking the required detection signature should withhold operating power instead.</p><p>Before swapping injectors, check the exact endpoint model and hardware revision against the injector’s voltage, powered pairs, polarity and current capability. Keep passive injectors labeled with their electrical specifications and intended equipment; an unidentified spare is a poor troubleshooting tool.</p><h3>Myth 2. Forcing power on is just another troubleshooting step</h3><p>A port that refuses to supply power may be rejecting an incompatible connection. Overriding detection removes the protection that prevents damage.</p><p>MikroTik documents that its RouterOS forced-on mode disables resistance detection and applies power regardless of whether a compatible device is detected. Power remains enabled after the cable is unplugged. Someone subsequently patching another device into that connection can therefore encounter an already energized circuit.</p><p>Short-circuit and overload protection remain active in this mode, but neither establishes that the endpoint accepts the supplied voltage or polarity. Overcurrent protection and compatibility detection serve different purposes.</p><p>Use forced power only for a verified combination of source, endpoint and wiring. Document the exception at the port and patching location. Do not assume another vendor’s “static” or “manual” power setting behaves identically; check whether it reserves a power budget, changes detection behavior or actually energizes an otherwise unrecognized connection.</p><h3>Myth 3. A compliant splitter makes any DC-powered device safe</h3><p>A PoE splitter introduces a second electrical interface that the upstream switch cannot validate. Its PoE input may comply with IEEE 802.3af, while its DC output may be completely wrong for the equipment attached to it.</p><p>For example, TP-Link’s TL-POE10R accepts standards-compliant PoE and provides selectable 5-, 9-, or 12-volt DC output. The switch detects and powers the splitter. It does not determine whether the device connected to the splitter’s barrel plug requires 5 volts or 12 volts.</p><p>Selecting 12 volts for equipment rated only for 5 volts can damage that equipment, even if PoE connects successfully upstream. TP-Link explicitly warns that an incorrect splitter output voltage can damage hardware.</p><p>Match output voltage, connector dimensions and polarity to the endpoint, then verify sufficient output current at the selected voltage. A plug that fits establishes only mechanical compatibility. Label adjustable splitters with the intended output setting before placing them into service.</p><h3>Myth 4. A working data link proves the cabling can handle PoE</h3><p>A link that carries Ethernet successfully does not prove the cabling can carry the intended DC load under actual installation conditions. Thermal concerns increase with higher-power PoE, particularly Type 3 and Type 4 deployments in large or tightly packed cable bundles.</p><p>For a given resistance, resistive heating increases with the square of the current. Dense bundles also dissipate heat less effectively than isolated runs. The risk depends on actual current, bundle configuration, ambient temperature, conductor gauge, and the cable’s temperature rating. Replacing an access point does not automatically create a thermal hazard.</p><p>Higher temperatures increase insertion loss, which can cause data errors before physical damage occurs. Sustained excessive heat can also prematurely age cable jacketing materials, causing irreversible deterioration. Siemon identifies this aging mechanism in its guidance on remote-powering installations. A link that recovers when it cools does not prove that prolonged operation at excessive temperatures is harmless.</p><p>Before deploying higher-power endpoints over existing cabling, assess the installed bundle and pathway against the expected load. Verify conductor specifications and temperature ratings, inspect terminations and test DC resistance and resistance unbalance. These electrical tests complement the thermal assessment; they do not prove that a dense bundle will stay within its temperature rating under load.</p><h3>Myth 5. Unplugging a powered connection causes no wear beyond ordinary patching</h3><p>Disconnecting an RJ45 connection while it carries PoE can produce an arc as the contacts separate. The resulting erosion can damage the plated contact surfaces.</p><p>Siemon’s technical explanation distinguishes the arcing location from the final contact position when the plug is fully seated. Damage at the working contact surface can create an unreliable connection and increase bit errors. Connector geometry therefore matters: some designs move the arcing region away from the surfaces used during normal operation.</p><p>This does not mean every live disconnection destroys a connector. It means repeated disconnection under load is both an electrical endurance issue and a mechanical one.</p><p>For planned repatching, disable PoE on the relevant port before disconnecting. When selecting connecting hardware, look for documented testing under IEC 60512–99–001 or IEC 60512–99–002, as appropriate to the intended application. These standards address connector endurance during separation under electrical load; verify that the documented test conditions cover the intended PoE use.</p><p>Inspect suspect plugs and jacks when intermittent faults follow repeated maintenance. Replacing only the patch cord may leave a damaged jack in service.</p><h3>Myth 6. PoE protection also covers outdoor surges</h3><p>Device detection and power limiting manage normal PoE operation. They do not establish that an outdoor copper run is adequately protected against transient voltage entering along the Ethernet cable.</p><p>Axis explains that nearby electrical and magnetic fields can induce voltage transients in network cabling. These can reach a camera or propagate toward indoor network equipment. Built-in surge resistance has limits and depends on correct installation. This exposure also affects Ethernet connections that do not carry PoE.</p><p>Protection requires attention to the complete path. In the shielded installation Axis describes, shielding must remain continuous through cables, couplings and patch panels to a properly grounded power source. A shielded patch cord attached to equipment without the required ground path does not complete that protection.</p><p>Where the installation design requires it, use PoE-compatible surge protectors at the building entry and near a distant outdoor endpoint, and install grounding according to the manufacturer’s instructions. Verify support for the intended PoE type and data rate. The switch’s ability to detect a camera and supply its normal operating power says nothing about whether that outdoor cable can deliver a damaging surge to the switch.</p><h3>The next replacement is the real test</h3><p>These failures rarely start with the original installation. They start when the person who set up the system has moved on. A technician replaces a failed 24V passive injector with the 48V unit on the shelf. Someone patches a new device into a port left on forced power. Someone else swaps a shielded patch cord on an outdoor run for an unshielded one, breaking the shield path. Each change looks routine, and each one reverses a decision nobody recorded. Put those decisions where the next person will see them: on the port label, in the switch port description and on the injector itself. A PoE installation is only compatible if it stays compatible through the next repair.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst3750/software/troubleshooting/g_power_over_ethernet.html">Cisco: PoE detection, power allocation and overcurrent protection</a></li><li><a href="https://ethernetalliance.org/blog/2018/02/06/the-hazards-of-passive-poe/">Ethernet Alliance: The hazards of passive PoE</a></li><li><a href="https://manual.mikrotik.com/docs/hardware/poe-out/">MikroTik: PoE modes and electrical compatibility</a></li><li><a href="https://www.tp-link.com/us/business-networking/accessory/tl-poe10r/v5/">TP-Link: TL-POE10R splitter specifications</a></li><li><a href="https://support.omadanetworks.com/uk/document/122252/">Omada: PoE installation and splitter voltage warnings</a></li><li><a href="https://www.flukenetworks.com/expertise/learn-about/four-pair-power-over-ethernet-poe-and-your-cabling-plant">Fluke Networks: Four-pair PoE and cabling requirements</a></li><li><a href="https://www.siemon.com/en/environments/smart-buildings/power-over-ethernet-poe/">Siemon: PoE contact erosion and thermal aging of cabling</a></li><li><a href="https://www.ul.com/services/poe-connector-testing-iec-60512-99-001-and-iec-60512-99-002">UL Solutions: PoE connector endurance testing</a></li><li><a href="https://whitepapers.axis.com/en-us/power-surges">Axis: Power surges and network equipment protection</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=77761d2def29" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[8 tips for surviving firmware upgrades on production networks]]></title>
            <link>https://medium.com/@networktigers/8-tips-for-surviving-firmware-upgrades-on-production-networks-d9b0e17ead5e?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/d9b0e17ead5e</guid>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[upgrade]]></category>
            <category><![CDATA[firmware-update]]></category>
            <category><![CDATA[firmware]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Sat, 31 Oct 2026 19:26:01 GMT</pubDate>
            <atom:updated>2026-10-31T19:26:01.542Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*GjG6FdiQqDN5xd6e" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>Successful firmware upgrades depend less on the upgrade itself than on preserving a safe path forward and a reliable path back. First published September 29, 2026, on </em><a href="https://news.networktigers.com/network-knowhow/tips-for-surviving-firmware-upgrades-on-production-networks/"><em>NetworkTigers News</em></a><em>.</em></p><p>Firmware upgrades on production networks are rarely difficult because of the install command. The risk comes from what happens between the old version and the new one: temporary mixed-version states, hidden firmware dependencies, failed reboots and rollback paths that are less straightforward than expected.</p><h3>Tip 1: Map the intermediate states, not just the target version</h3><p>An upgrade plan that only records the current and target versions misses much of the operational risk.</p><p>Some platforms require intermediate releases. Others update ROMMON, bootloaders or other firmware as part of the process.</p><p>Document those intermediate states before the window starts. Identify which device will be forwarding traffic, which components will be unavailable and whether the vendor supports each temporary combination. This also gives a more realistic maintenance window than relying on the advertised reboot time.</p><h3>Tip 2: Treat mixed software versions as a separate operating state</h3><p>Sequential upgrades keep redundant systems available, but they also create periods when peers are running different code.</p><p>That can affect synchronization, state replication, stack membership and failover behavior. Redundancy can also hide <a href="https://news.networktigers.com/network-knowhow/when-redundancy-makes-the-outage-worse/">shared dependencies and failover behavior that only become apparent under stress</a>.</p><p>Check what the surviving device must handle while its peer is offline, then check what happens when that peer returns on the new version. Forwarding ownership, LAG behavior, routing state and synchronization matter more here than whether both devices eventually show as healthy.</p><h3>Tip 3: Set the upgrade order from network dependencies</h3><p>Rack order is rarely the safest upgrade order.</p><p>Management access, authentication, routing, uplinks and failover paths can all depend on devices elsewhere in the same change. Rebooting one switch can move traffic onto another device that is due to be upgraded next, or remove the management path needed to recover a later failure.</p><p>Build the sequence around those dependencies. The right order is the one that keeps forwarding, management and recovery paths available for the next step.</p><h3>Tip 4: Find the point where rollback becomes recovery</h3><p>Rollback is not always as simple as reinstalling the previous image.</p><p>Firmware upgrades can also change bootloaders, ROMMON versions, configuration formats or other components that are not automatically reversed when the network operating system is downgraded. Some downgrade paths also require their own installation process.</p><p>Before the upgrade, establish what rollback actually restores and where in the sequence a simple reversal stops being possible. After that point, the fallback plan may be a separate recovery procedure rather than a return to the original state.</p><h3>Tip 5: Assume older hardware can fail at its next reboot</h3><p>Long uptime does not prove that a device will boot cleanly.</p><p>A switch can forward traffic for years without exercising its full startup path. An upgrade forces it to read images from flash, initialize hardware, run POST and bring every component back from a cold state. Aging flash, power supplies or other marginal hardware can fail at that point.</p><p>For older or long-running equipment, have <a href="https://news.networktigers.com/hardware-hub/minimum-hardware-requirements-small-business-network/">spare hardware and current configuration backups</a> ready before the reboot. A spare only helps if you can quickly put the correct hardware, modules, software, and configuration into service.</p><h3>Tip 6: Define the conditions that stop the upgrade</h3><p>Maintenance windows create pressure to keep moving, even when the network behaves unexpectedly.</p><p>A device may return with delayed routing convergence, incomplete synchronization or intermittent packet loss. If you upgrade the next device before you understand the problem, the change becomes harder to diagnose and the blast radius grows. That becomes particularly costly when <a href="https://news.networktigers.com/all-articles/the-real-reason-network-outages-take-so-long-to-diagnose/">network visibility starts disappearing at the same time as the fault</a>.</p><p>Set stop conditions before the work starts. They should be based on observable behavior such as failed HA synchronization, unexpected route changes, packet loss, interfaces returning in the wrong state or a device exceeding its expected recovery time.</p><h3>Tip 7: Validate traffic paths, not just device health</h3><p>An upgraded device can be reachable, show healthy hardware and have every important interface up while production traffic is still affected.</p><p>Validation should follow the services that actually cross the device. Depending on its role, that can include routing adjacencies, LACP members, spanning-tree state, DHCP forwarding, 802.1X authentication, tunnels, multicast or application traffic.</p><p>Compare those results with the pre-upgrade state before moving to the next device. Management reachability proves that the device came back. It does not prove that the network is behaving normally.</p><h3>Tip 8: Leave time for delayed problems to appear</h3><p>Do not fill the entire maintenance window with upgrade activity.</p><p>Routing protocols need time to reconverge. Redundant systems resynchronize. Endpoints renew leases and authentication sessions. Monitoring systems clear transient alerts. Some problems only become visible when sessions reconnect or traffic shifts back to its normal path.</p><p>Set a latest-start time for the final upgrade step. If earlier work consumes that margin, stop rather than finishing the last device with no time left to observe the network.</p><h3>Firmware upgrades are about controlling the transition</h3><p>The vendor can document supported versions and upgrade procedures, but it cannot account for the dependencies, hardware condition and traffic paths in a production network. Those factors determine whether an unexpected result stays contained or becomes an outage.</p><p>They need to be worked out before the maintenance window. Once an upgrade is underway, there is less time to investigate dependencies, arrange replacement hardware or work out what rollback actually requires.</p><h3>Sources</h3><ul><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-9300-series-switches/222280-upgrading-catalyst-9300-switches.html">Cisco: Upgrade Catalyst 9300 Switches</a></li><li><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst_standalones/b-in-service-software-upgrade-issu.html">Cisco: In-Service Software Upgrade (ISSU)</a></li><li><a href="https://www.juniper.net/documentation/us/en/software/junos/high-availability/topics/concept/nssu-ex-series.html">Juniper Networks: Understanding Nonstop Software Upgrade on EX Series Switches</a></li><li><a href="https://arubanetworking.hpe.com/techdocs/AOS-CX/Consolidated_RNs/6300-6400%20Release%20Notes/GUID-CF623A2D-986A-4BF5-A6DC-65DD4B57A1EA.html">HPE Aruba Networking: AOS-CX 6300/6400 upgrade information</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About NetworkTigers. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d9b0e17ead5e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The credibility bias of the factory seal]]></title>
            <link>https://grain.saltindex.com/the-credibility-bias-of-the-factory-seal-6840290ef068?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/6840290ef068</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[network-security]]></category>
            <category><![CDATA[network-reliability]]></category>
            <category><![CDATA[network-equipment]]></category>
            <category><![CDATA[hardware]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Thu, 29 Oct 2026 19:22:35 GMT</pubDate>
            <atom:updated>2026-11-02T10:27:54.750Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*Tfa0jEZTnHBt3d6k" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>Unproven hardware stays off the suspect list too long. New hardware gets a grace period it has not earned. First published June 19, 2026, on </em><a href="https://news.networktigers.com/hardware-hub/the-credibility-bias-of-the-factory-seal/"><em>NetworkTigers News</em></a><em>.</em></p><p>When service degrades after a cutover, troubleshooting instincts often move past the physical layer too quickly. The first suspects are often configuration templates, routing protocols, MTU mismatches, firewall policies, firmware behavior, carrier handoffs, and missed migration steps. Those are reasonable places to look. They are also what gives bad hardware room to hide.</p><h3>Signs of life are not proof of function</h3><p>Total failure is easy to diagnose. A module that refuses to power on, a port that stays dark, or a device that will not boot gives the team a straight diagnostic path. The real nightmares are caused by components that work just enough to stay believable.</p><p>An optic can negotiate a link and pass basic light-level checks, yet still drop frames under production load. A line card can boot clean, accept its configuration, and pass ICMP traffic, then fail under specific ASIC-level forwarding patterns. A power supply can bring a chassis online and handle a quiet control plane, then go electrically unstable the moment the system is under actual demand. A network adapter can appear clean to the operating system while behaving badly when firmware, driver behavior, throughput, and heat all arrive at once.</p><p>Drops get attributed to provider congestion. One-way traffic gets attributed to asymmetric routing. Flapping sessions get attributed to a firewall. Poor throughput gets attributed to buffering or MTU. Each explanation is technically plausible. A component can show signs of life and still be functionally dead.</p><h3>Complexity provides cover</h3><p>A weekend window can include new switches, upgraded optics, rewritten firewall policies, firmware changes, and a new carrier handoff all happening simultaneously. When the environment changes in several places at once, the investigation gravitates toward the interactions between those changes. A bad optic or a flaky cable looks too crude to explain a nuanced, intermittent routing issue.</p><p>The newly installed component sits at the center of the fault domain, but the surrounding complexity makes it easier to treat as part of the intended design rather than part of the problem. The same dynamic appears in <a href="https://news.networktigers.com/hardware-hub/the-lifecycle-trap-when-still-working-is-not-good-enough">hardware lifecycle decisions</a>, where an aging platform can feel safer than the replacement that is supposed to succeed it.</p><h3>Replacement parts must earn their status</h3><p>The answer is not to assume the new component is bad. It is to avoid treating it as known-good before it has earned that status. Staging does not prove a component will survive production load, but it catches failures that should not reach the maintenance window. A part that has been powered, recognized, and checked against the expected platform, code, or firmware enters the window differently than one pulled straight from a box or shelf.</p><p>During a change, the same logic applies. The more variables introduced at once, the less clean the evidence becomes. When the window allows it, prove the replacement against a known baseline before adding new behavior. If the fault appears after the baseline holds, the investigation starts in a different place than it would if hardware, firmware, topology, and policy all changed at the same time.</p><p>Comparison still matters. If the symptoms fit a physical fault, move the suspect part, replace it with a known-good equivalent, clear the counters, and see whether the fault follows the component or stays with the path. That is not random swapping. It is fault isolation.</p><p>Factory-new, seller-refurbished, and spare hardware are unproven until tested. A test record or clean history can reduce uncertainty. Neither proves the component will work in this environment. Test new components soon after acquisition, not months later during a cutover or outage. Early testing catches obvious failures. The rest can only be proved in stages: during staging, at cutover, and finally under the weight of actual production traffic.</p><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6840290ef068" width="1" height="1" alt=""><hr><p><a href="https://grain.saltindex.com/the-credibility-bias-of-the-factory-seal-6840290ef068">The credibility bias of the factory seal</a> was originally published in <a href="https://grain.saltindex.com/">Salt Index</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why on-prem infrastructure is making a quiet return]]></title>
            <link>https://medium.com/@networktigers/why-on-prem-infrastructure-is-making-a-quiet-return-e91fde6af49b?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/e91fde6af49b</guid>
            <category><![CDATA[network-design]]></category>
            <category><![CDATA[network-architecture]]></category>
            <category><![CDATA[repatriation]]></category>
            <category><![CDATA[on-prem]]></category>
            <category><![CDATA[network-infrastructure]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Tue, 27 Oct 2026 17:46:52 GMT</pubDate>
            <atom:updated>2026-10-27T17:46:52.969Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*GtgVU1nYs-MnJwgf" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><em>Remote visibility only goes so far when the infrastructure itself fails and recovery requires direct control. First published September 25, 2026, on </em><a href="https://news.networktigers.com/cloud-chronicles/why-on-prem-infrastructure-is-making-a-quiet-return/"><em>NetworkTigers News</em></a><em>.</em></p><p>Some infrastructure teams are moving selected systems back on-prem, and cost is only one consideration. <a href="https://news.networktigers.com/cloud-chronicles/fallback-plan-every-business-needs-when-the-cloud-goes-down/">When something fails</a>, they want the problem to remain within infrastructure they can inspect and act on.</p><p>Cloud infrastructure removed a great deal of routine work. In doing so, it also moved parts of the system beyond the customer’s reach. That is often an acceptable trade. Managed services let relatively small teams run systems that once required far more people, equipment, and capital.</p><p>The tradeoff becomes harder to accept when the infrastructure itself is the problem.</p><p>When an application starts returning errors, authentication becomes unreliable, and requests time out, engineers work through the layers they own but eventually reach a point where the evidence stops.</p><h3>When on-prem visibility matters</h3><p>Infrastructure abstraction helps by hiding implementation details. Customers do not need to know which physical host runs a VM or where an object is stored. During an outage, however, that hidden layer may contain the answer.</p><p>An engineer working on on-prem infrastructure can inspect the network, storage, hypervisor, and recent configuration changes. More importantly, the team has authority to act on what it finds. It can isolate a failed component, reverse a change, or move a workload.</p><p>While these actions do not guarantee a quick recovery, they do generate evidence and reduce uncertainty.</p><h3>Visibility does not always mean control</h3><p>Visibility and control are not the same thing. A team can have telemetry showing that a managed service is failing and still have no access to the layer causing the failure or any authority to repair it. It can also own infrastructure and instrument it badly. The advantage appears when ownership gives engineers both evidence and a way to act on it.</p><p>This limitation also changes incident command. A team that suspects its own network can assign an owner, gather telemetry, make a reversible change, and test the result. A team that suspects a provider often has to open a support case and wait while continuing to rule out internal causes.</p><p>That slows recovery. Engineers become reluctant to make aggressive changes when they cannot tell whether they are correcting the incident or creating a second one.</p><p>Google Cloud’s June 12, 2025 outage demonstrated the consequences of this boundary:</p><p>According to Google’s incident report, a policy change containing unintended blank fields was written into regional Spanner tables used by Service Control. The data propagated globally within seconds. Service Control binaries encountered the unexpected values and entered crash loops.</p><p>The same mechanism that distributes state across Google’s systems also carried the bad configuration across regions. Regional separation therefore did not isolate customers from the failure.</p><p>Google’s engineers could identify the bad policy and repair the affected systems, but customers could not. There was no server to reboot or policy object they could fix. The failure sat outside the customer’s remediation boundary.</p><h3>Hidden dependencies make redundancy hard to prove</h3><p>The same day produced another example.</p><p>Cloudflare reported a major outage affecting Workers KV after infrastructure it depended on at a third-party cloud provider became unavailable. Because other Cloudflare products depended on KV, the outage spread into Access, WARP, Workers AI, and other services.</p><p>This is the harder problem in modern cloud architecture: customers rarely know the full dependency graph.</p><p>A company may buy services from several vendors and believe it has diversified, but those vendors may still rely on common infrastructure underneath.</p><p>That makes resilience difficult to assess from the outside. Multi-vendor is not automatically multi-failure-domain, and customers cannot always see where supposedly separate services converge.</p><p>With infrastructure the organization owns and operates, it can verify whether redundant systems follow separate physical and logical paths, rather than assuming service separation equals failure-domain separation.</p><p>That does not make the design resilient by itself. It does, however, make hidden coupling easier to find before an outage exposes it in production.</p><h3>On-prem only works if the team can operate it</h3><p>Private infrastructure is not inherently safer.</p><p>Uptime Institute’s 2026 outage analysis found that serious data-center outages remained expensive, with 57% of respondents who had experienced a major outage reporting costs above $100,000 and one in five reporting more than $1 million. Power remained a leading cause of impactful outages, and failure to follow established procedures remained a major source of human-error-related incidents.</p><p>Moving workloads on-prem also moves responsibility for those risks, which can be a poor decision for organizations without enough infrastructure depth.</p><p>A server room with weak cooling, undocumented switching, untested failover, or <a href="https://news.networktigers.com/network-knowhow/when-tribal-knowledge-becomes-a-single-point-of-failure/">only one administrator who understands the storage platform</a> is not more resilient just because the hardware is nearby.</p><p>Direct control requires <a href="https://news.networktigers.com/featured/network-documentation-and-why-it-matters/">current documentation</a>, useful telemetry, tested recovery procedures, controlled configuration, and people who understand how the environment behaves when components fail.</p><p>Without that, reclaiming ownership only increases the number of things the team itself can break without actually improving recovery.</p><h3>Repatriation is a placement decision</h3><p>Flexera’s 2025 State of the Cloud research found that respondents had moved about 21% of cloud workloads back to private or data-center environments while continuing to migrate other workloads into public cloud. Its 2026 research again showed widespread hybrid environments rather than a broad reversal.</p><p>The available repatriation data does not show enterprises abandoning public cloud at scale. Instead, it indicates that infrastructure teams are making placement decisions workload by workload rather than treating the cloud as the default destination.</p><p>A stable internal service with predictable demand has different requirements from an application with bursty global traffic. The same applies when one workload can tolerate provider-side recovery timelines and another cannot. Defaulting to “cloud-first” becomes less useful than asking where a particular workload is easiest to operate and recover, what it costs to run there, and how much control the team needs when it fails.</p><p>World Wide Technology’s 2026 analysis of repatriation identifies loss of operational control as one reason a workload may no longer fit the public-cloud model. Its analysis points to constraints such as provider-controlled maintenance, service changes, hardware choices, and platform limits that reduce the customer’s control over how and when the underlying environment changes. For workloads where those constraints affect recovery or performance, placement becomes an operational decision rather than simply a hosting preference.</p><p>That does not require abandoning the operating practices cloud adoption helped normalize. Infrastructure as code, automated deployment, containers, and modern telemetry work on owned infrastructure too. Moving a workload on-prem does not have to mean returning to manually managed servers.</p><h3>Where should the diagnostic boundary sit?</h3><p>Cost is easy to compare because it produces numbers. Operational control is harder to price. Its value becomes visible during a serious outage.</p><p>How far can the team follow the failure before it reaches infrastructure someone else controls? If the provider confirms the incident but offers no workaround, does the customer still have a meaningful recovery action available?</p><p>The same review should expose less obvious dependencies. A workload moved onto owned hardware has not necessarily regained operational independence if its critical control plane still depends on outside services the team cannot inspect or repair.</p><p>The useful decision, then, is not how much infrastructure an organization can bring back in-house. It is which failures it is willing to outsource and where it wants its engineers’ authority to end.</p><p>For some workloads, that loss of control is an acceptable price for managed infrastructure. For others, it is becoming a reason to move the diagnostic boundary closer to the engineers responsible for keeping the service running.</p><p>That is the practical appeal behind the quiet return of on-prem. When a service fails, the team responsible for restoring it can keep following the evidence instead of stopping at a provider boundary. They may still have a difficult outage to solve, but they have access to the failed system and the authority to do something about it.</p><p>For the workloads where that matters, knowing exactly what broke is worth owning more of what can break.</p><h3>Sources</h3><ul><li><a href="https://status.cloud.google.com/incidents/ow5i3PPK96RduMcb1SsW">Google Cloud — “Multiple GCP products are experiencing Service issues,” June 12, 2025 incident report</a></li><li><a href="https://blog.cloudflare.com/cloudflare-service-outage-june-12-2025/">Cloudflare — Cloudflare service outage, June 12, 2025</a></li><li><a href="https://intelligence.uptimeinstitute.com/resource/annual-outage-analysis-2026">Uptime Institute — Annual Outage Analysis 2026</a></li><li><a href="https://info.flexera.com/CM-REPORT-State-of-the-Cloud">Flexera — 2026 State of the Cloud Report</a></li><li><a href="https://www.flexera.com/blog/finops/the-latest-cloud-computing-trends-flexera-2025-state-of-the-cloud-report/">Flexera — 2025 State of the Cloud Report</a></li><li><a href="https://www.wwt.com/blog/from-cloud-first-to-cloud-right-repatriation-without-regret-5-of-7">World Wide Technology — From Cloud-First to Cloud-Right: Repatriation without Regret, May 14, 2026</a></li></ul><p>*****</p><p>By Ben Walker, a <a href="https://www.networktigers.com/">NetworkTigers</a> Author.</p><p>Ben Walker is a freelance research-based technical writer. He has worked as a content QA analyst for AT&amp;T and Pernod Ricard.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e91fde6af49b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network switch configuration: 7 checks before going live]]></title>
            <link>https://medium.com/@networktigers/network-switch-configuration-7-checks-before-going-live-309362208b76?source=rss-47f1d5f569dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/309362208b76</guid>
            <category><![CDATA[configuration-management]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[network-design]]></category>
            <category><![CDATA[network-equipment]]></category>
            <category><![CDATA[switch]]></category>
            <dc:creator><![CDATA[NetworkTigers]]></dc:creator>
            <pubDate>Mon, 26 Oct 2026 21:29:48 GMT</pubDate>
            <atom:updated>2026-10-26T21:29:48.933Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/436/0*PSLM86xj1CEPikel" /><figcaption>Image licensed from Dreamstime.com</figcaption></figure><p><strong>Editor’s Note:</strong> This article was originally published on September 25, 2021 on on <a href="https://news.networktigers.com/network-knowhow/network-switch-configuration-top-tips-and-practices/">NetworkTigers News</a>. It was substantially rewritten and updated in October 2026 to correct technical inaccuracies and expand the guidance on switch configuration and deployment.</p><p><em>Network switch configuration determines which VLANs ports belong to, what traffic uplinks carry and who can administer the device.</em></p><p>Configuring a network switch means defining how it connects devices and handles their traffic. An employee’s workstation might connect through an access port assigned to one VLAN, while an uplink carries traffic for several VLANs to another switch. Separate settings control administrative access and how redundant links behave.</p><p>Before deployment, verify that these settings match the intended network design and work together as expected. Commands, defaults and feature behavior vary by switch model and software release.</p><h3>1. Establish the starting configuration before connecting uplinks</h3><p>Stage the switch away from production. Record its model, software version, installed modules and intended role. Compare the existing configuration with the deployment plan rather than assuming that a spare, refurbished unit or factory-default device is ready to connect.</p><p>Resolve inherited settings before deployment: management addresses, local accounts, authentication servers, VLANs, trunks and routing configuration. For Cisco environments using VLAN Trunking Protocol, check the version, mode, domain and VLAN database before joining the switch to the network. A revision number alone is not a complete deployment check.</p><p>Use the manufacturer’s reset procedure when you need a clean start. Deleting a configuration file is not a universal reset method. Complete any necessary software changes during staging, with a documented <a href="https://news.networktigers.com/network-knowhow/tips-for-surviving-firmware-upgrades-on-production-networks/">firmware upgrade and recovery plan</a>.</p><h3>2. Prove that management access is restricted</h3><p>Identify which administrative hosts should reach the switch and enforce that boundary. A dedicated management VLAN provides logical separation, but its name does not prevent routed access from other networks. Apply access controls at the appropriate management interface or network boundary.</p><p>Use SSH for command-line administration and disable unnecessary management services. On Cisco IOS and IOS XE, enabling SSH does not by itself exclude Telnet: VTY transport settings determine which remote terminal protocols are accepted. Check every applicable VTY line.</p><p>Test from both an authorized administrative host and a user network that should be denied. A successful administrator login proves that only the permitted path works. Keep a working console or independent recovery path available while changing remote access controls.</p><h3>3. Test authentication failure before it becomes an outage</h3><p>Verify that administrators receive the intended privileges, rather than assigning unrestricted access to every account. When using centralized authentication, authorization, and accounting, confirm the switch can reach the servers and that administrative activity is recorded.</p><p>During staging, test the approved recovery procedure with the authentication service unavailable. Do not assume that a local account will automatically provide access. Fallback behavior depends on the configured authentication methods and the failure condition; a server rejecting credentials is different from a server not responding.</p><p>Document who can retrieve emergency credentials and how use of those credentials is audited. Recovery that depends on an unavailable colleague is not ready for an overnight incident.</p><h3>4. Verify VLAN membership at both ends of each link</h3><p>Assign each port a documented purpose: endpoint access, an explicitly designed endpoint trunk, a switch uplink or a routed interface. Check the operational state and the intended configuration.</p><p>For trunks, compare allowed VLANs and native VLAN settings at both ends. On Cisco Catalyst platforms covered by the cited guide, trunks allow all VLANs by default. Limit that list to the VLANs the link actually needs.</p><p>A physical link can stay up even if one side’s allowed list is missing a VLAN. Native VLAN mismatches can also place untagged traffic into different VLANs at opposite ends. Test a representative endpoint in each required VLAN rather than relying on a ping to the switch’s management address.</p><p>Verify segmentation separately. VLAN membership does not establish which traffic routing policies or firewalls permit between those VLANs.</p><h3>5. Apply loop protection according to the port’s role</h3><p>Confirm the intended spanning-tree root and expected forwarding paths before adding redundant links. Investigate any unexpected root or blocked uplink before the switch carries production traffic.</p><p>Use edge-port behavior, such as Cisco PortFast, where the attached device and topology justify it. Do not apply an endpoint template indiscriminately to inter-switch links.</p><p>BPDU guard can disable an edge port that receives a spanning-tree bridge protocol data unit. That protects the topology, but also creates an operational obligation: monitor the event and investigate the connection before restoring service.</p><p>BPDU filtering is not a substitute. Cisco warns that interface-level filtering can effectively disable spanning tree on that interface and allow loops. Administratively disable unused ports so an undocumented connection does not immediately become an active network path.</p><h3>6. Test redundancy with a failure, not a diagram</h3><p>If the design includes redundant uplinks or a port channel, verify the actual forwarding state and member participation. Two connected cables do not prove that both belong to the intended bundle or that the surviving path has enough capacity.</p><p>During an approved test window, remove one intended failure component at a time and observe application connectivity, interface counters and recovery behavior. Define acceptable interruption before testing. Include management access in the check: forwarding may recover while the administrative path remains unavailable.</p><p>Account forport-channel <a href="https://news.networktigers.com/network-knowhow/6-times-a-port-channel-is-the-wrong-solution/">limits</a>. A bundle can survive a member failure while leaving too little bandwidth for the workload.</p><h3>7. Verify persistence, monitoring and restoration</h3><p>Save or commit the accepted configuration using the platform’s procedure. On Cisco IOS and IOS XE, changing the running configuration does not automatically save those changes to the startup configuration.</p><p>Keep a protected backup outside the device. Record the corresponding software version and restoration procedure, including any dependencies a configuration file alone cannot supply.</p><p>Confirm that logs reach the collector with usable timestamps and that monitoring detects a deliberately generated test event. Check interface errors and discards against a recorded baseline. A configured logging destination is not proof that anyone receives the messages.</p><p>Where feasible, restart the switch during staging and repeat the acceptance checks. This tests the saved configuration and boot behavior before production depends on them.</p><h3>Make the next change easier to verify</h3><p>Keep the acceptance results with the configuration: permitted management sources, required VLAN paths, expected spanning-tree state and observed failover behavior. Those records give the next engineer something concrete to compare after a replacement, upgrade or unexplained outage. A deployment checklist becomes substantially more useful when it preserves evidence of what passed.</p><h3>Sources</h3><ul><li><a href="https://sec.cloudapps.cisco.com/security/center/resources/ios_xe_integrity_assurance.html">Cisco: IOS XE software integrity assurance and administrative controls</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/security-vpn/secure-shell-ssh/4145-ssh.html">Cisco: SSH configuration and access restrictions</a></li><li><a href="https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/sec_usr_aaa/configuration/xe-16/sec-usr-aaa-xe-16-book/sec-cfg-authentifcn.html">Cisco: Authentication methods and fallback behavior</a></li><li><a href="https://www.cisco.com/c/en/us/support/docs/switches/catalyst-2940-series-switches/109304-manage-vlandat.html">Cisco: Managing the VLAN database on Catalyst switches</a></li><li><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9300/software/release/17-10/configuration_guide/vlan/b_1710_vlan_9300_cg/configuring_vlan_trunks.html">Cisco Catalyst 9300: VLAN trunk configuration</a></li><li><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9300/software/release/17-3/configuration_guide/lyr2/b_173_lyr2_9300_cg/configuring_optional_spanning_tree_features.html">Cisco Catalyst 9300: PortFast, BPDU guard and BPDU filtering</a></li></ul><p>*****</p><p>By Katrina Boydon: Katrina is a veteran technology writer and editor known for turning complex ideas into clear, readable insights. She embraces AI as a helpful tool but keeps the editing, and the skepticism, firmly human.</p><p>About <a href="https://www.networktigers.com/">NetworkTigers</a>. Founded in 1996, NetworkTigers sells new and used network equipment routers, switches and firewalls and has a special interest and expertise in cybersecurity. Find out more at <a href="https://www.networktigers.com/">NetworkTigers.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=309362208b76" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>