<?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 Ajay Shankar BP on Medium]]></title>
        <description><![CDATA[Stories by Ajay Shankar BP on Medium]]></description>
        <link>https://medium.com/@ajayshankar17.bp?source=rss-184be75785f4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*F6t9sxad-_Zfh9fA</url>
            <title>Stories by Ajay Shankar BP on Medium</title>
            <link>https://medium.com/@ajayshankar17.bp?source=rss-184be75785f4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 19:27:03 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ajayshankar17.bp/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[AWS Multi-Party Approval (MPA): Stopping the “One Person, One Mistake” Problem in Your Cloud]]></title>
            <link>https://medium.com/@ajayshankar17.bp/aws-multi-party-approval-mpa-stopping-the-one-person-one-mistake-problem-in-your-cloud-c85f4d797c9f?source=rss-184be75785f4------2</link>
            <guid isPermaLink="false">https://medium.com/p/c85f4d797c9f</guid>
            <category><![CDATA[aws-security]]></category>
            <category><![CDATA[cloud-security]]></category>
            <category><![CDATA[zero-trust]]></category>
            <category><![CDATA[multi-party-approval]]></category>
            <category><![CDATA[aws-organizations]]></category>
            <dc:creator><![CDATA[Ajay Shankar BP]]></dc:creator>
            <pubDate>Fri, 17 Jul 2026 09:17:10 GMT</pubDate>
            <atom:updated>2026-07-17T09:17:10.628Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>How AWS is implementing team-based, Zero Trust decision-making in your most sensitive operations</em></blockquote><h3>The Problem: What Happens Without Approval?</h3><blockquote><strong><em>One identity = one decision.</em></strong></blockquote><p>AWS IAM follows a straightforward authorization model; if an identity has permission to act, AWS executes that action immediately. There is no native approval workflow, secondary authorization, or mandatory peer review before the request is processed.</p><p>While this model provides operational simplicity, it also introduces significant security and governance challenges for organizations managing critical cloud environments.</p><h3>Risks of Single-Step Authorization</h3><p>When a single IAM identity can perform sensitive operations independently, the following risks arise:</p><ul><li><strong>Single point of failure — </strong>one mistake, one bad click, one irreversible action (deleted vault, exported key, wiped resource)</li><li><strong>Insider threat — </strong>a malicious or coerced insider can act completely alone</li><li><strong>Compromised credentials — </strong>a stolen session inherits full, unchecked power instantly</li><li><strong>Compliance gaps — </strong>regulated industries need documented multi-person sign-off; plain IAM can’t enforce that</li><li><strong>Backups share the same blast radius — </strong>if prod credentials are compromised, your “safety net” backup can be destroyed too</li></ul><h3>The Missing Control</h3><p>Traditional security controls, such as logging, monitoring, and auditing, are primarily<strong> detective controls; </strong>they help identify what happened after an action has already occurred.<br>What is often missing is a <strong>preventive approval mechanism</strong> that requires a second authorized individual to approve high-risk operations before AWS executes them.</p><h3>The Solution: What Is AWS Multi-Party Approval</h3><p>AWS Multi-party approval (MPA) is a capability of AWS Organizations that protects a predefined list of sensitive operations behind a distributed, team-based approval process. Instead of one identity being able to execute a protected action unilaterally, MPA requires a configurable number of approvers (an “M of N” threshold) to sign off before the operation is allowed to run.</p><figure><img alt="In short: MPA turns a single point of failure into a team decision." src="https://cdn-images-1.medium.com/max/1024/1*hCF-D00u9ZNUJrrWIBxMvg.png" /><figcaption>In short: MPA turns a single point of failure into a team decision.</figcaption></figure><h3>A Quick Word on Zero Trust</h3><h4>You’ll hear “Zero Trust” a lot in this context, so it’s worth a beat.</h4><p>Zero Trust is a security model built on one core idea: <strong>never trust, always verify, </strong>regardless of whether a request comes from inside or outside your network, and regardless of who is making it. Instead of assuming that an authenticated identity is automatically safe to act on, every sensitive request is independently verified, every single time.</p><blockquote>Classic IAM answers the question <em>“Is this identity allowed to do this?” </em>Zero Trust pushes further and asks, <em>“Should this specific action, right now, actually be allowed to happen — even if the identity is valid?”</em></blockquote><p>This is precisely where MPA fits in. Valid credentials and correct permissions are no longer treated as sufficient proof of intent for high-risk operations. A second (or third) independent human has to actively verify the request before it executes. That’s Zero Trust applied not just to authentication, but to the <em>action</em> itself — which is what makes MPA such a natural fit for protecting things like backup recovery and cryptographic key trust anchors.</p><h3>The Three Key Personas</h3><p>MPA’s workflow is built around three distinct roles, each with a clear responsibility:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IrS6v0rgtmxpo2YdbUVz9A.png" /></figure><p><strong>1. Requester: </strong>The individual or entity attempting to execute a protected operation. The moment they trigger that action, an approval session begins, and they wait for the team’s decision.</p><p><strong>2. Administrator: </strong>Responsible for building and maintaining approval teams. The admin creates a team, sets the approval threshold, and invites approvers. Once a team is active, it becomes self-protecting, and even the admin can’t unilaterally change team membership or thresholds without going through the same approval process.</p><p><strong>3. Approver: </strong>The people who actually review and vote on requests. When a team is active, approvers get an email notification for every pending request, review the details in the MPA portal, and choose to approve or reject.</p><p>This separation of duties is the whole point: the person who <em>asks</em> isn’t the same as the person who <em>decides</em>, and the person who <em>manages the team</em> can’t quietly override the team either.</p><h3><strong>How It Works: The Three-Step Approval Process</strong></h3><ul><li><strong>Step 1 — Request:</strong> requester triggers a protected operation → it enters <strong>pending</strong> → approvers are notified by email.</li><li><strong>Step 2 — Approval:</strong> approvers vote <strong>Approve</strong> or <strong>Reject</strong> in the MPA portal within <strong>24 hours</strong>. No response = rejection.</li><li><strong>Step 3 — Result:</strong> threshold met → <strong>auto-executed</strong> with the requester’s permissions. Threshold missed or expired → <strong>rejected</strong>, nothing runs.</li></ul><h3>How to Configure MPA with AWS Backup</h3><p>The flagship use case for MPA today is protecting <strong>logically air-gapped vaults</strong> in AWS Backup, isolated, tamper-resistant backup storage designed to survive a compromise of your primary account. Here’s how the setup flows end to end.</p><h3>Step 1: Set Up Multi-Party Approval (Organization Management Account)</h3><ol><li>Sign in to your AWS Organizations management account.</li><li>Open the Organizations console → Multi-party approval.</li><li>Choose Set up Multi-party approval.</li><li>MPA will detect (or prompt you to create) your IAM Identity Center instance.</li><li>Choose Complete setup.</li></ol><h3>Step 2: Create an Approval Team</h3><ol><li>In the MPA console, choose Create team.</li><li>Provide a Name and Description.</li><li>Choose Add approvers and select at least 3 IAM Identity Center users.</li><li>Set the minimum required approvals to at least 2 (recommended: below the total approver count).</li><li>Choose Create team — invitations go out by email, and the team becomes active once every invited approver accepts.</li></ol><h3>Step 3: Create a Logically Air-Gapped Vault in AWS Backup</h3><p>Create your vault as usual in the AWS Backup console, noting its ARN — you’ll need it in the next step.</p><h3>Step 4: Associate the Approval Team with the Vault</h3><ol><li>In the AWS Backup console, go to Backup vaults and select your logically air-gapped vault.</li><li>On the vault details page, choose Assign approval team.</li><li>Select the approval team you created, optionally add a comment, and choose Send request.</li></ol><h3>Step 5: Requesting Recovery Access</h3><p>When disaster strikes, a requester in the recovery account asks for temporary access to the vault, which kicks off a fresh MPA approval session.</p><p>Approvers get notified, review the request in the MPA portal, and vote. Once the threshold is met, AWS Backup automatically creates a restore access backup vault in the requester’s account — giving them a controlled, time-boxed window into the air-gapped backups without ever having direct standing access to them.</p><h3>Closing Thoughts</h3><p>Multi-party approval is AWS’s answer to a problem that IAM alone was never designed to solve: making sure that for your <em>most</em> sensitive operations, no single identity, human, or compromised can act alone. By formalizing the requester/administrator/approver relationship and layering it on top of your existing permissions model, MPA turns high-stakes actions into genuine team decisions, with an audit trail to prove it.</p><p>If you’re running AWS Backup with logically air-gapped vaults or handling payment cryptography key imports, MPA is one of the more meaningful <strong>“set it up once, sleep better forever”</strong> controls available today.</p><p><em>Have you implemented MPA in your organization? I’d love to hear how you’ve structured your approval teams — drop a comment below.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c85f4d797c9f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Real-Time Detection of Risky Firewall Rules in OCI Network Firewall]]></title>
            <link>https://medium.com/@ajayshankar17.bp/real-time-detection-of-risky-firewall-rules-in-oci-network-firewall-335bff5ebb3f?source=rss-184be75785f4------2</link>
            <guid isPermaLink="false">https://medium.com/p/335bff5ebb3f</guid>
            <category><![CDATA[firewall-security]]></category>
            <category><![CDATA[oci-network-firewall]]></category>
            <category><![CDATA[oci]]></category>
            <dc:creator><![CDATA[Ajay Shankar BP]]></dc:creator>
            <pubDate>Sun, 03 May 2026 15:56:51 GMT</pubDate>
            <atom:updated>2026-05-03T16:33:38.038Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>A practical guide to building real-time OCI Network Firewall rule validation using Events, Functions, and Notifications — so every risky rule is flagged the moment it is saved, before it becomes a problem.</blockquote><p>OCI Network Firewall lets you define rules that control traffic between your subnets, VCNs, and the internet. Each rule has three key components: a source (where the traffic originates), a destination (where it is directed), and a port or protocol (the type of traffic permitted).</p><p>Let’s consider a real-world scenario. You’re a security engineer managing workloads in OCI. A developer urgently needs access to a VM and updates the Network Firewall policy, allowing source: ANY, destination: ANY, port: ANY, action: ALLOW can be saved and activated without any warning. The firewall is updated. Nobody is notified. The risk is invisible until someone reviews the policy — or until an incident reveals it.</p><p>This is not an unusual scenario. It plays out in cloud environments of every size and maturity. However, a deeper problem emerges when you zoom out to the team level.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s7rk1NRmr2lfHhst5cEJhg.png" /></figure><blockquote><strong>Note :</strong> There is no native “rule review gate.” When a firewall policy update is saved, it takes effect immediately,no built-in warning, no validation check, no alert if 0.0.0.0/0 was just allowed .That visibility gap is exactly what this article addresses.</blockquote><h3>The solution: event-driven rule validation</h3><p>To solve this, I built a custom alerting and visibility framework using three native OCI services like OCI Events, OCI Functions, and OCI Notifications. The system triggers automatically whenever a firewall policy is updated, validates every rule in the policy against a defined set of security conditions, and sends an immediate alert if anything looks risky. No servers to manage, no scheduled jobs, no manual steps involved.</p><h3><strong>Architecture Overview</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Jz4OrfTGM6LZsz0pELQOZg.png" /></figure><h3>How it works — concept</h3><ul><li>Every time a firewall policy is updated in OCI, the platform automatically emits an audit event called UpdateNetworkFirewallPolicy</li><li>An OCI Events rule is configured to listen for this specific event type and invoke the OCI Function whenever it fires</li><li>The function does not receive rule details inside the event payload; it uses the OCI Network Firewall API to fetch the complete, current policy</li><li>Every rule in the policy is then evaluated against a predefined set of security conditions</li><li>If any rule triggers a violation, the function publishes a formatted alert to the OCI Notifications topic, and the subscribed email address receives it within Minutes</li></ul><blockquote><strong><em>Remember: </em></strong>The OCI function always fetches and validates the full policy, not just the rule that changed. This means every update cycle also re-checks all existing rules, so nothing slips through undetected between reviews.</blockquote><h3>Implementation steps</h3><ol><li><strong>Create the Notifications topic and email subscription</strong></li></ol><p>Navigate to <strong>OCI Console → Developer Services → Notifications</strong></p><ul><li>Click <strong>Create Topic</strong> — name it something descriptive, for example firewall-policy-alerts</li><li>Once the topic is created, open it and click <strong>Create Subscription</strong></li><li>Set the protocol to <strong>Email</strong> and enter the security engineer’s email address</li><li>OCI will send a confirmation email to that address — click the link inside it to activate the subscription</li><li>Copy the <strong>Topic OCID</strong> — the function in Step 2 will need this to publish alert messages</li></ul><p><strong>2. Function Code :</strong></p><p>Git repo: https://github.com/ajaybpshankar/OCI-NetworkFirewall-Rule-Review</p><p><strong>3. Create the OCI Events rule</strong></p><p>Navigate to <strong>OCI Console → Observability &amp; Management → Events Service</strong></p><ul><li>Click <strong>Create Rule</strong></li><li>Under <strong>Rule Conditions</strong>, configure the following:</li><li>Service Name: Network Firewall</li><li>Event Type: Update Network Firewall Policy</li><li>Under <strong>Actions</strong>, select <strong>Functions</strong></li><li>Choose the function application and the function you created in Step 2</li><li>Click <strong>Save Rule</strong></li></ul><p><em>A firewall is only as strong as the process behind it. When validation is automated, every change is a reviewed change, and every engineer on the team becomes accountable, even without anyone watching over their shoulder.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=335bff5ebb3f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>