<?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 aravind.mutthe on Medium]]></title>
        <description><![CDATA[Stories by aravind.mutthe on Medium]]></description>
        <link>https://medium.com/@aravind.mutthe?source=rss-19aa7ffc5f3e------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*jlFntbUpMEY8JFhVk-u-QA@2x.jpeg</url>
            <title>Stories by aravind.mutthe on Medium</title>
            <link>https://medium.com/@aravind.mutthe?source=rss-19aa7ffc5f3e------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 11:52:58 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@aravind.mutthe/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[CONTROL PLANE (The “Orchestrator”)]]></title>
            <link>https://medium.com/@aravind.mutthe/control-plane-the-orchestrator-2351a3a542e3?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/2351a3a542e3</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Wed, 29 Apr 2026 03:50:08 GMT</pubDate>
            <atom:updated>2026-04-29T04:10:31.849Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Jub6SHVJ_rrcKr-HRHDbjg.jpeg" /><figcaption>Sy</figcaption></figure><ol><li><strong>CONTROL PLANE (The “Orchestrator”)</strong></li></ol><p><strong>Pipes that configure, signal, and manage:</strong></p><p>├ <strong>Netlink sockets </strong>(kernel ↔ daemon signaling ,ip, bridge, route cmds)<br>├ <strong>Docker/containerd socket </strong>(/var/run/docker.sock, docker ps, run, stop)<br>├ <strong>Kubernetes API </strong>(kube-apiserver, kubectl apply, delete)<br>├ <strong>systemd D-Bus</strong> (service control, systemctl start, stop)<br>├ <strong>SSH control channel</strong> (port forwarding setup)</p><p>├ <strong>RabbitMQ control queues</strong> (exchanges, bindings)<br>└ <strong>TLS handshake </strong>(certificate negotiation)</p><p><strong>Commands</strong>: <strong>ip, docker, kubectl, systemctl, ssh -L</strong></p><p><strong>2. DATA PLANE (The “Carrier”)</strong></p><p><strong>Pipes that move actual user/application data:</strong></p><p>├── <strong>HTTP/HTTPS</strong> (web pages, API responses)<br>├── <strong>SSH data channel</strong> (shell output, file transfers)<br>├── <strong>WebSockets</strong> (real-time messages)<br>├── <strong>gRPC streams</strong> (protobuf payloads)<br>├── <strong>Message buses</strong> (Kafka records, RabbitMQ msgs)<br>├── <strong>TCP/UDP sockets</strong> (raw data streams)<br>├── <strong>Container veth pairs</strong> (container nw /container traffic (veth + bridge))<br>└── <strong>Unix domain sockets</strong> (local IPC data)</p><p><strong>Commands: curl, ssh, wscat, grpcurl, kafka-console-consumer</strong></p><p><strong>3. MANAGEMENT PLANE (The “Observer”)</strong></p><p><strong>Pipes that expose metrics, logs, and state:</strong></p><p>├── <strong>procfs </strong>(/proc) (CPU, memory, processes)<br>├── <strong>sysfs </strong>(/sys) (hardware sensors, devices)<br>├── <strong>cgroupfs </strong>(/sys/fs/cgroup) (container resource limits)<br>├── <strong>Application logs</strong> (/var/log/, journald)<br>├── <strong>Trace pipes</strong> (strace, bpftrace, OpenTelemetry, Jaeger)</p><p>├── <strong>Container metrics</strong> (docker stats, cAdvisor)<br>└── <strong>Network statistics</strong> (/proc/net/dev, ss, netstat, tcpdump)</p><p><strong>Commands: cat /proc/meminfo, journalctl, top, iostat</strong></p><p><strong>4. HARDWARE PLANE (The “Physical”)</strong></p><p><strong>Pipes that interface with physical hardware:</strong></p><p>├── <strong>Network interfaces</strong> (eth0, wlan0)</p><p>├── <strong>Character devices </strong>(/dev/tty*, /dev/input/*)</p><p>├── <strong>Hardware drivers</strong> (in kernel)<br>├── <strong>Block devices</strong> (/dev/sda, /dev/nvme0n1)<br>├── <strong>GPIO sysfs interface</strong> (/sys/class/gpio/,(embedded))<br>├── <strong>I2C/SPI devices</strong> (/dev/i2c-*, /dev/spidev*)<br>├── <strong>PCI device files</strong> (/proc/bus/pci/*)<br>└── <strong>USB device files</strong> (/dev/bus/usb/*)</p><p><strong>Commands: dmesg, lsblk, i2cget, gpio read,dmesg -w, ls -la /dev/, lspci -v<br> cat /sys/class/thermal/thermal_zone0/temp, hdparm -t /dev/sda</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2351a3a542e3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Pipes/Sockets]]></title>
            <link>https://medium.com/@aravind.mutthe/network-pipes-sockets-92ebb17166cc?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/92ebb17166cc</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Wed, 29 Apr 2026 03:37:41 GMT</pubDate>
            <atom:updated>2026-05-10T02:04:41.596Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iX4nq_HFM96Uyv4VAQQHAQ.jpeg" /></figure><h3>To master these network “pipes,” you need to use the iproute2 suite and standard network diagnostic tools. Here is the complete command list for their entire lifecycle.</h3><h3><strong>🛰️ 1. TCP &amp; UDP (Transport Pipes)</strong></h3><p>These pipes are <strong>ephemeral</strong>; they exist as long as the process is running.</p><p><strong>Creation </strong>(Lifecycle Start):</p><p>nc -l [port] (Create a TCP listener/pipe)</p><p>nc -u -l [port] (Create a UDP listener/pipe)</p><p><strong>Management </strong>&amp; Inspection:</p><p>ss -tulpn (List all active pipes: TCP, UDP, Listening, Process ID, Numeric)</p><p>lsof -i :[port] (Find exactly which User/Account owns a specific pipe)</p><p><strong>Control </strong>(Configuration):</p><p>sysctl net.ipv4.tcp_window_scaling=1 (Adjust pipe “width” or buffer)</p><p>ulimit -n (View max number of pipes a user can open)</p><p><strong>Termination</strong>:</p><p>kill -9 [PID] (Force close the pipe by killing the owner process)</p><h3>🧥 2. VPN &amp; GRE Tunnels (Virtual Interface Pipes)</h3><p>These are <strong>persistent </strong>pipes. They stay “up” even if no data is flowing.</p><p>GRE Tunnels</p><p><strong>Creation</strong>:</p><p>ip tunnel add gre1 mode gre remote [Remote_IP] local [Local_IP]</p><p>ip link set gre1 up (Activate the pipe)</p><p><strong>Configuration</strong>:</p><p>ip addr add 10.0.0.1/24 dev gre1 (Assign an internal IP to the pipe)</p><p><strong>Management</strong>:</p><p>ip -s tunnel show gre1 (Show Stats: how much data flowed through)</p><p><strong>Deletion</strong>:</p><p>ip tunnel del gre1</p><p>WireGuard (Modern VPN Pipe)</p><p><strong>Creation</strong>:</p><p>wg genkey | tee privatekey | wg pubkey &gt; publickey (Generate “Locks” for the pipe)</p><p>wg-quick up wg0 (Start the pipe using a config file)</p><p><strong>Management</strong>:</p><p>wg show (See “Handshake” status and data flow)</p><p>Control:</p><p>systemctl enable wg-quick@wg0 (Systemd control to start pipe at boot)</p><h3>💀 3. Raw Sockets (Hardware Hooks)</h3><p>These do not have a standard “lifecycle” because they are instantaneous <strong>hooks </strong>into the network card.</p><p><strong>Usage (Capture):</strong></p><p>tcpdump -i eth0 (Open a raw pipe to see everything on the wire)</p><p><strong>Usage (Injection):</strong></p><p>hping3 -S [IP] -p 80 (Inject custom packets into a raw pipe)</p><p><strong>Security Control:</strong></p><p>setcap ‘cap_net_raw+ep’ /path/to/program (Give a specific program permission to use raw pipes without being Root)</p><h3>🛠️ Management &amp; Control Summary Table</h3><p>Action -&gt; TCP/UDP — -Tunnel (GRE/VPN) — -Raw Sockets</p><p>Creation-&gt; socket() / nc ip link / wg-quick socket(AF_PACKET)</p><p>View Status-&gt; ss -atn ip addr / wg show netstat -i</p><p>Owner-&gt; App Process Root / Systemd Root / UID 0</p><p>Persistence	Dies with Process	Stays until ip del	Instantaneous</p><p>Logging	/var/log/syslog	journalctl -u wg0	tcpdump</p><h4><strong>📋 The “Grand Command” for Troubleshooting</strong></h4><p>If you ever lose track of these pipes, use this command to see everything (IPCs and Network):</p><p><strong>sudo lsof -i -n -P | grep -E ‘TCP|UDP|raw’</strong></p><p>-i: Shows Internet files (sockets).</p><p>-n: Don’t resolve hostnames (faster).</p><p>-P: Don’t resolve port names (shows 80 instead of http).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=92ebb17166cc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Pipes/Sockets for IPC]]></title>
            <link>https://medium.com/@aravind.mutthe/network-pipes-sockets-for-ipc-3eef4933a9a0?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/3eef4933a9a0</guid>
            <category><![CDATA[ipcc]]></category>
            <category><![CDATA[linux]]></category>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Sun, 19 Apr 2026 18:04:37 GMT</pubDate>
            <atom:updated>2026-05-10T02:04:30.021Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iX4nq_HFM96Uyv4VAQQHAQ.jpeg" /></figure><p>Socket pipe=A socket is a foundational endpoint for two-way communication in computer networking, They are essential for enabling network communication (e.g., client-server) and providing secure electrical connections</p><h3>🛰️ 1. TCP Sockets</h3><p>Relationship: Strictly Client-Server.</p><p>It connects two unrelated processes on different machines.</p><p>Storage (The “File”): Stored in the Kernel’s Socket Table (not on disk).</p><p>You can view it at /proc/net/tcp.</p><p>Owner/Account:</p><p><strong>The Listener</strong>: Usually a System Daemon (e.g., sshd runs as root, nginx runs as www-data).</p><p><strong>The Connector:</strong> Usually a User Process (e.g., your browser running as your local username).</p><p><strong>Who Creates it:</strong> The Kernel creates the socket when a program calls socket().</p><p><strong>What flows:</strong> Raw Bytes. The data is put there by the User App, but the Kernel adds “signals” (Sequence numbers, ACKs).</p><p><strong>Command</strong>: ss -at (Shows all TCP pipes and their owners).</p><h3><strong>🏹 2. UDP (The Datagram Pipe)</strong></h3><p>Relationship: Distributed/Peer-to-Peer.</p><p>There is no “connection,” just a mailbox.</p><p><strong>Storage</strong>: Stored as a Port Binding in the Kernel’s RAM. Viewable at /proc/net/udp.</p><p><strong>Owner/Account:</strong> Often used by System Services like systemd-resolved (runs as systemd-resolve user) for DNS queries.</p><p><strong>Who Creates it</strong>: The Kernel, triggered by an app calling bind() to a port.</p><p><strong>What flows:</strong> Discrete Packets. The User Process puts the data in; the Kernel just attaches an IP header and dumps it to the wire.</p><p><strong>Command</strong>: ss -au (Shows all UDP mailboxes).</p><h3><strong>🧥 3. GRE &amp; VPN (The Virtual Tunnel Pipes)</strong></h3><p>Relationship: Site-to-Site. It links two separate networks.</p><p><strong>Storage</strong>: This is the only one that feels like a “file.” It is stored as a Virtual Network Interface in the Kernel’s Device Tree.</p><p><strong>Owner/Account: </strong>Managed by Systemd or the Root User.</p><p>Who Creates it:</p><p><strong>GRE</strong>: The Kernel (via the ip command).</p><p><strong>VPN</strong>: A Daemon (e.g., wireguard or openvpn running as root or a dedicated vpn service account).</p><p><strong>What flows</strong>: Encapsulated IP Traffic. The Kernel Routing Engine puts data into this pipe automatically when a packet is destined for the remote network.</p><p><strong>Command</strong>: ip link show (Look for tun0, wg0, or gre0).</p><h3><strong>💀 4. Raw Sockets (The Hardware Pipe)</strong></h3><p>Relationship: Direct Hardware Access.</p><p><strong>Storage</strong>: It has no name or port. It is a direct “hook” into the Network Card (NIC) Driver.</p><p><strong>Owner/Account:</strong> Strictly Root/UID 0. Standard users cannot create these for security reasons.</p><p><strong>Who Creates it:</strong> The Kernel, but only if the process has CAP_NET_RAW privileges.</p><p><strong>What flows:</strong> Raw Ethernet/IP Frames. The User Process puts the entire packet — headers and all — into the pipe.</p><p><strong>Command:</strong> tcpdump -i any (This tool creates a raw socket to “sniff” the wire).</p><h3>📊 Comparison of Network “Pipes” vs. IPC</h3><p>Feature IPC -&gt;(Named Pipe)- Network Pipe (TCP) — Tunnel Pipe (VPN/GRE)</p><p>Storage-&gt; Name Path on Disk (/tmp/p)-IP:Port Pair (In RAM)-Interface Name (tun0)</p><p>Creator-&gt; mkfifo (User)- socket() (App)-ip link (Root/Daemon)</p><p>Manager -&gt;Kernel VFS -Kernel Network Stack — Kernel + Systemd Daemon</p><p>Account-&gt; File Owner (User) -Process Owner (www-data) -Root / System Service</p><p>Data Flow-&gt; RAM to RAM- RAM → NIC → Wire → RAM -RAM → Encryption → Wire</p><h4><strong>⚓ Key Difference in Management</strong></h4><p><strong>IPC Pipes:</strong> Managed by the Virtual File System (VFS). The kernel treats it like a file that never hits the disk.</p><p><strong>Network Pipes:</strong> Managed by the Network Stack. The kernel treats it like a “State Machine” that has to track if the other side of the world is still there.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3eef4933a9a0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[1. Types of “Pipes” for Communication]]></title>
            <link>https://medium.com/@aravind.mutthe/types-of-pipes-for-communication-76bda43c6976?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/76bda43c6976</guid>
            <category><![CDATA[ci-cd-pipeline]]></category>
            <category><![CDATA[socket-programming]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[named-pipes]]></category>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Sun, 19 Apr 2026 17:39:57 GMT</pubDate>
            <atom:updated>2026-05-02T12:25:30.605Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Jub6SHVJ_rrcKr-HRHDbjg.jpeg" /></figure><p><strong>Pipe</strong>= a powerful IPC mechanism that allowing data to flow from one process to another through a virtual channel</p><p>In computer systems &amp; networking, we have several “pipes” depending on what layers or scope we talk about. Let’s map them clearly.</p><h3><strong>1. IPC pipes-B/w the processes on the same Machine</strong></h3><ol><li><strong>Anonymous pipes</strong> → one-way, for one task</li></ol><p>used by related process or parent child processes.</p><p>temporary channels (used in shells: ls | grep foo).</p><p>unidirectional channel, empheral</p><p>2. <strong>Named pipes (FIFOs)</strong> → permanent file-like objects,</p><p>mkfifo mypipename=Name stored on disk and data on Kernal Buffer(RAM)</p><p>It’s a special file, stored on disk. persistant</p><p>unidirectional,</p><p>repeated tasks or multiple tasks.</p><p>Named Pipes provide built-in synchronization — the reader automatically waits for the writer</p><p>3. <strong>Unix domain sockets (/tmp/socket)</strong> → client-server/demon- service</p><p>faster, file-system path instead of IP:port.</p><p>like TCP sockets, but local only,</p><p>4.<strong>Message queues / shared memory</strong> → for distributed or bg worker processes</p><p>higher-level IPC, fast, used by OS subsystems.</p><p>These are like indoor plumbing pipes inside the same house (one computer).</p><p>Named pipes are generally faster than network sockets for local IPC because they bypass the entire network stack (no packetization, checksums, or routing).</p><p>Stack Overflow</p><h4>Typical Performance (Latency)</h4><h4>Data Storage — Best Use Case</h4><p>Shared Memory Fastest (sub-microsecond) -RAM -Massive data sets (GBs)</p><p>Named Pipes Fast (~5–30 ms)- Kernel Buffer- Streaming logs, shell scripts</p><p>Unix Sockets Fast (~25–45 ms) - Kernel Buffer- Local web servers, databases</p><p>TCP Sockets Moderate (~40–60+ ms) -Network Stack -Cross-machine co</p><p>Key Insight: While Shared Memory is technically faster, it requires manual synchronization (mutexes/semaphores). Named Pipes provide built-in synchronization — the reader automatically waits for the writer.</p><h3><strong>2. Network pipes (between devices)</strong></h3><p>•	TCP sockets (stream, reliable, ordered).</p><p>•	UDP sockets (datagram, fast, no guarantees).</p><p>•	Raw sockets (direct control of IP packets).</p><p>•	QUIC streams (UDP-based, encrypted, modern replacement for TCP in some web use cases, e.g., HTTP/3).</p><p>•	VPN tunnels (IPSec, WireGuard, OpenVPN) → encapsulate traffic in a secure “pipe” between networks.</p><p>•	GRE tunnels / MPLS / SD-WAN tunnels → provider-level pipes for routing customer traffic.</p><p>👉 These are like underground city pipes connecting houses (devices).</p><h3>3. Application-level “pipes”</h3><p>•	HTTP/HTTPS → Web traffic inside TCP/QUIC sockets.</p><p>•	SSH → Secure remote shell over TCP (port 22).</p><p>•	TLS → Encryption layer that wraps traffic in a secure pipe.</p><p>•	WebSockets → Bidirectional pipe over HTTP.</p><p>•	gRPC / REST / SOAP APIs → App-level communication “pipes” built over HTTP/TCP.</p><p>•	Message buses → Kafka, RabbitMQ, MQTT (IoT) → pipes for events/messages.</p><p>👉 These are like specialized transport pipes for different materials (water, gas, electricity = web, shell, APIs, IoT).</p><h3>4. System/kernel pipes (netlink -demon)</h3><p>•	Netlink sockets → kernel ↔ user space communication in Linux (e.g., ip route).</p><p>•	Procfs/sysfs → pseudo-files that act like pipes from kernel info to user programs.</p><p>•	Character devices (/dev/null, /dev/tty) → streams to/from hardware or kernel.</p><p>👉 These are service ducts inside the building, carrying control signals, not just data.</p><p>⚡ Visual analogy</p><p>• <strong>Socket </strong>= garden hose → attaches to spigot (IP+port) and lets water (data) flow.</p><p>• Pipe types above = water mains, indoor plumbing, gas lines, electricity conduits → each one is a different infrastructure for carrying something.</p><p>•	HTTPS = clean drinking water flowing inside the hose (encrypted, safe).</p><h3>Quick cheat list (finger-tips)</h3><p>• <strong>Local only</strong>: Pipes, FIFOs, Unix</p><p>sockets, shared memory.</p><p>• <strong>Network</strong>: TCP, UDP, raw sockets, QUIC, VPN, tunnels.</p><p>• <strong>Application</strong>: HTTPS, SSH, WebSockets, APIs, message queues.</p><p>• <strong>System</strong>: Netlink, procfs, device pipes.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=76bda43c6976" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[1. Between processes on the same machine (IPC pipes)]]></title>
            <link>https://medium.com/@aravind.mutthe/1-between-processes-on-the-same-machine-ipc-pipes-8bc343d47eab?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/8bc343d47eab</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Sat, 18 Apr 2026 15:32:24 GMT</pubDate>
            <atom:updated>2026-05-10T02:04:18.898Z</atom:updated>
            <content:encoded><![CDATA[<p>IPC pipe is a simple memory buffer</p><p>•	Anonymous pipes → one-way, temporary channels (used in shells: ls | grep foo).</p><p>•	Named pipes (FIFOs) → permanent file-like objects (mkfifo mypipe).</p><p>•	Unix domain sockets (/tmp/socket) → like TCP sockets, but local only, faster, file-system path instead of IP:port.</p><p>•	Message queues / shared memory → higher-level IPC, fast, used by OS subsystems.</p><p>👉 These are like indoor plumbing pipes inside the same house (one computer).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8bc343d47eab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Internal Pipes or IPC]]></title>
            <link>https://medium.com/@aravind.mutthe/internal-pipes-or-ipc-31df0cd8f216?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/31df0cd8f216</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Sat, 18 Apr 2026 15:17:47 GMT</pubDate>
            <atom:updated>2026-05-10T02:03:56.074Z</atom:updated>
            <content:encoded><![CDATA[<p>👥 1. <strong>Anonymous Pipe for commands</strong></p><p>Relationship: Parent and Child processes only (e.g., a shell and the command it runs).</p><p>Where it’s stored:</p><p>Name: Nowhere. It is “anonymous” and has no file entry.</p><p>Data: Strictly in Kernel RAM.</p><p>How it’s created: Created automatically by the shell or via the pipe() system call in code.</p><p>Who uses it: Processes with a parent-child relationship.</p><p>Who controls it: The Kernel. It exists only in RAM and handles the synchronization.</p><p>Commands:</p><p>Operation: The | symbol.</p><p>Example: ls | grep “txt” (The kernel creates the pipe to link the two).</p><p>🏷️<strong> 2. Named Pipe (FIFO) for scripts</strong></p><p>Relationship: Independent Processes on the same machine (they don’t need to be related).</p><p>Where it’s stored:</p><p>Name: A special file in the Filesystem (e.g., /tmp/my_pipe).</p><p>Data: Strictly in Kernel RAM (the file on disk is always 0 bytes).</p><p>How it’s created: Using the mkfifo command.</p><p>Who uses it: Independent processes on the same machine.</p><p>Who controls it: The Kernel (data) and Filesystem (the name).</p><p>Commands:</p><p>Create: mkfifo /tmp/my_pipe</p><p>Write: echo “data” &gt; /tmp/my_pipe</p><p>Read: cat &lt; /tmp/my_pipe</p><p>Delete: rm /tmp/my_pipe</p><p><strong>🧦 3. Unix Domain Socket (Demon and services/process) for web servers</strong></p><p>Relationship: Client and Server processes on the same machine.</p><p>Where it’s stored:</p><p>Name: A “socket file” in the Filesystem (e.g., /run/docker.sock).</p><p>Data: Strictly in Kernel RAM.</p><p>How it’s created: By a service (like Docker or MySQL) using the socket() system call.</p><p>Who uses it: System Services (Daemons) and their Clients.</p><p>Who controls it: Systemd often manages the “socket file,” but the Kernel moves the data.</p><p>Example (Systemd Socket Activation):</p><p>Service: Docker or MySQL.</p><p>Configuration: A .socket file in /etc/systemd/system/.</p><p>Commands:</p><p>Identify: ls -l (Look for an s at the start, like srwxr-xr-x).</p><p>Connect: nc -U /path/to/socket (using Netcat).</p><p>Monitor: ss -x (Shows all active Unix sockets).</p><p>📥<strong> 4. Message Queue (POSIX) (Distributed systems or bg workers)</strong></p><p>Relationship: Producer and Consumer (Independent processes; supports “Fire and Forget”).</p><p>Where it’s stored:</p><p>Name: A virtual path in a Kernel-managed namespace (e.g., /my_queue).</p><p>Data: In Kernel RAM, but it persists even if no one is currently reading it.</p><p>How it’s created: Using mq_open() in C or libraries in Python/Go.</p><p>Who uses it: Distributed apps or background workers (Producers/Consumers).</p><p>Who controls it: The Kernel (for POSIX) or a Service (like Redis/RabbitMQ).</p><p>Example (Redis as a Message Queue):</p><p>Commands:</p><p>List: ls /dev/mqueue (on Linux, the kernel mounts the queue here).</p><p>Inspect: ipcs -q (Shows current message queues and usage).</p><p>Remove: ipcrm -q &lt;ID&gt;</p><p>📊 Quick Summary Table</p><p>IPC Type	Used By	Stored (Name)	Stored (Data)	Creation Cmd</p><p>Anon Pipe	Parent/Child	Nowhere	RAM	`</p><p>Named Pipe	Unrelated	Filesystem	RAM	mkfifo</p><p>Unix Socket	Client/Server	Filesystem	RAM	socket() call</p><p>Msg Queue	Producer/Cons	Kernel Mem	RAM	mq_open()</p><p>⚓ Key Takeaway</p><p>Pipes are for streaming data (like water).</p><p>Sockets are for bidirectional talk (like a phone call).</p><p>Queues are for messages (like a mailbox).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=31df0cd8f216" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[System Architecture Planes]]></title>
            <link>https://medium.com/@aravind.mutthe/system-architecture-planes-linux-server-view-20ed6f0f2cd1?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/20ed6f0f2cd1</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 13:54:54 GMT</pubDate>
            <atom:updated>2026-04-19T18:46:29.759Z</atom:updated>
            <content:encoded><![CDATA[<h3>Linux Server View</h3><p>We’ll divide everything into 5 <strong>practical planes</strong>:</p><ol><li><strong>Control Plane</strong> →Process orchestration, lifecycle, policy</li><li><strong>Data Plane</strong> → actual work (serving data, processing)</li><li><strong>Management Plane</strong> → admin access &amp; observability</li><li><strong>Security Plane</strong> → enforcement &amp; protection</li><li>Network Plane-.</li></ol><p>Each plane has:</p><ul><li><strong>Tightly coupled core users /services/demons</strong></li><li><strong>Loosely coupled external services/plugins to install with package mangers yum,apt,rpm</strong></li></ul><h3>1. CONTROL PLANE (The Brain)</h3><p>In a server environment, the Control Plane ensures <strong>High Availability (HA)</strong>. If the control plane freezes, the server becomes a “zombie” — it might still serve web pages, but you can’t stop, start, or fix anything.</p><h3>🔒 Tightly Coupled demons (Core)</h3><ul><li><strong>systemd:</strong> Acts as the <strong>Process Supervisor</strong>. In servers, it uses “Watchdogs.” If a service hangs, systemd notices the heartbeat failure and forcibly restarts it.</li><li><strong>systemd-logind:</strong> On a server, this manages your SSH “Session State.” It ensures that when you log out, background processes don’t accidentally get killed (unless specified).</li><li><strong>systemd-sysusers/tmpfiles:</strong> These ensure the <strong>Idempotency</strong> of the OS. Every time you boot, the server checks that the required system users and /run sockets exist. This is the &quot;Self-Healing&quot; aspect of the core.</li></ul><p>👤 Users:</p><ul><li>root</li><li>systemd-*</li></ul><p>👉 These define:</p><ul><li>Boot sequence</li><li>Service dependencies</li><li>Process supervision</li></ul><h3>Loosely Coupled (Optional)</h3><ul><li>cron → scheduled execution</li><li>systemd timers (hybrid model)</li><li>container runtimes (Docker, etc.)</li></ul><p>👤 Users:</p><ul><li>cron</li><li>docker</li></ul><p>👉 These extend control but don’t define the OS itself.</p><h3>2. DATA PLANE (The Workload)</h3><p>The “Golden Rule” of server architecture: <strong>The Core OS should be invisible.</strong> The Data Plane is where the “Value” is generated.</p><h3>🔒 Tightly Coupled Demons(Core)</h3><ul><li><strong>The Kernel:</strong> On a server, the kernel is tuned for <strong>Throughput</strong>. It handles the TCP stack and context switching.</li><li><strong>D-Bus:</strong> On servers, D-Bus is the “Internal API.” For example, if a RAID drive fails, the kernel tells the storage daemon, which tells the management plane via D-Bus.</li><li>This is where <strong>your server does its job</strong>.</li></ul><p>Very minimal here:</p><ul><li>Kernel (process execution, memory, I/O)</li><li>Basic IPC via D-Bus</li></ul><p>👉 Core OS does NOT serve data — it only enables it.</p><h3>🔌 Loosely Coupled (Main workload layer)</h3><h3>Web/App</h3><ul><li>NGINX</li><li>Apache HTTP Server</li><li>php-fpm</li></ul><h3>Database</h3><ul><li>PostgreSQL</li><li>MariaDB</li></ul><h3>Cache / queue</h3><ul><li>redis</li><li>memcached</li></ul><p>👤 Users:</p><ul><li>nginx</li><li>postgres</li><li>mysql</li><li>redis</li></ul><p>👉 This is fully <strong>loosely coupled by design</strong>:</p><ul><li>Communicates via TCP/Unix sockets</li><li>Horizontally scalable</li><li>Replaceable</li></ul><h3>3. MANAGEMENT PLANE (Observability)</h3><p>If you can’t measure it, you aren’t running a server; you’re running a hobby.</p><h3>🔒 Tightly Coupled (Core)</h3><ul><li><strong>systemd-journald:</strong> The <strong>Single Source of Truth</strong>. On a server, it is often configured to forward logs to a remote buffer immediately so that if a hacker deletes /var/log, the evidence is already gone to a safe location.</li><li><strong>systemd-coredump:</strong> Critical for <strong>DevOps</strong>. When a high-traffic C++ or Go binary crashes, this daemon saves the “stack trace” so developers can debug the exact memory state at the moment of failure.</li></ul><h3>🔒 Tightly Coupled demons (Core)</h3><ul><li>systemd-journald → logging backbone</li><li>systemd-coredump → crash analysis</li></ul><p>👤 Users:</p><ul><li>systemd-journal</li><li>systemd-coredump</li></ul><p>👉 Always present because:</p><ul><li>OS must log itself from boot</li></ul><h3>🔌 Loosely Coupled (Optional)</h3><h3>Remote access</h3><ul><li>OpenSSH (sshd)</li><li>cockpit (web UI)</li></ul><h3>Logging extensions</h3><ul><li>rsyslog</li><li>log forwarders</li></ul><h3>Monitoring</h3><ul><li>Prometheus exporters</li><li>Grafana agents</li></ul><p>👤 Users:</p><ul><li>sshd</li><li>cockpit</li><li>prometheus</li></ul><p>👉 These provide:</p><ul><li>Remote shell</li><li>Metrics</li><li>Dashboards</li></ul><h3>4. SECURITY PLANE (The Shield)</h3><p>Security is deep-layered (Defense in Depth).</p><h3>🔒 Tightly Coupled (Core)</h3><ul><li><strong>Sandboxing:</strong> Modern systemd service files use ProtectSystem=strict and PrivateTmp=yes. This means even if root is compromised in a specific service, the core OS files stay Read-Only.</li><li><strong>PAM (Pluggable Authentication Modules):</strong> This is the gatekeeper. Whether you use a password, an SSH key, or 2FA, PAM is the core logic that says “Yes” or “No.”</li></ul><h3>🔒 Tightly Coupled (Core)</h3><ul><li>Linux permissions (UID/GID model)</li><li>PAM (authentication framework)</li><li>systemd sandboxing</li></ul><p>👤 Users:</p><ul><li>root</li><li>nobody</li></ul><p>👉 This is baked into the OS kernel + base system.</p><h3>🔌 Loosely Coupled (Optional)</h3><h3>Firewall</h3><ul><li>UFW</li><li>iptables / nftables</li></ul><h3>Intrusion prevention</h3><ul><li>Fail2Ban</li></ul><h3>Encryption / access</h3><ul><li>SSH keys via OpenSSH</li></ul><p>👤 Users:</p><ul><li>fail2ban</li></ul><p>👉 These are:</p><ul><li>Policy-driven</li><li>Replaceable</li><li>External to core logic</li></ul><h3>5. NETWORK PLANE (The Gateway)</h3><p>In a server, the network is the only way in. Stability here is non-negotiable.</p><h3>🔒 Tightly Coupled</h3><ul><li><strong>systemd-networkd:</strong> The “Server Choice.” It is predictable. You define a .network file, and the IP is assigned. It doesn&#39;t try to &quot;scan for Wi-Fi&quot; or do anything fancy—it just locks the server onto the grid.</li><li><strong>systemd-resolved:</strong> Handles DNS caching. In a high-volume server (like a microservice), caching DNS locally saves milliseconds on every single database or API call, drastically increasing performance.</li></ul><h3>🔒 Tightly Coupled</h3><ul><li>systemd-networkd</li><li>systemd-resolved</li></ul><p>👤 Users:</p><ul><li>systemd-network</li><li>systemd-resolve</li></ul><h3>🔌 Loosely Coupled</h3><ul><li>NetworkManager</li><li>dhcpcd</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20ed6f0f2cd1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Core Linux Base OS]]></title>
            <link>https://medium.com/@aravind.mutthe/tightly-coupled-core-base-os-e51ba1fa74a5?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/e51ba1fa74a5</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 11:14:23 GMT</pubDate>
            <atom:updated>2026-04-19T18:39:10.173Z</atom:updated>
            <content:encoded><![CDATA[<h3>Tightly Coupled to os</h3><p>These are <strong>created automatically</strong> and are fundamental to system operation. Removing or breaking them can make the system unstable or unbootable.</p><ul><li>Core OS = <strong>state management + hardware abstraction</strong></li><li>Plugins = <strong>business logic + services</strong></li></ul><h3># Core system users (always present)</h3><p>These are <strong>system accounts (UID &lt; 1000)</strong>:</p><h3><strong>1. root</strong> UID0→ superuser</h3><p>unlimited authority on Linux system), owns the config files in /etc and hired another account</p><h3><strong>2. nobody</strong> →</h3><p>unprivileged fallback user, unlimited isolation.no home, no shell, no group.</p><p>key= none</p><p>nogroup</p><h3><strong>3</strong>. <strong>systemd-network -&gt;</strong> system connections</h3><p>Provides Isolation for network stack,</p><p>manages hardware(ips,links,routers), manages messy internet cables wifi/ethernet (wires and signals)</p><p>key they hold is (CAP_NET_ADMIN)= configure hardware</p><h3><strong>4. systemd-resolve</strong>-&gt; System names</h3><p>Provides isolation for dns process,</p><p>manages website names(dns,mDNS,LLMNR), but can&#39;t touch wires and your files.to handle messy internet (DNS).</p><p>handles untrusted dns names from the web.</p><p>key they hold is(CAP_NET_BIND_SERVICE)= listen to ports</p><h3><strong>5. systemd-timesync -&gt; System time</strong></h3><p>manages clock by talking NTP servers</p><p>talks to external clocks to stay accurate.</p><p>CAP_SYSTEM_TIME=set the sys clock</p><h3><strong>6. messagebus</strong> → messages</h3><p>used by D-Bus, it&#39;s a internal messaging system (IPC), that lets different programs/apps talks to each other. post office worker of the system</p><p>key = owns the D bus socket.</p><h3><strong>7. systemd-coredump</strong>-&gt; System crashes, forensic tech</h3><p>user job is to handle system crashes.</p><p>handles toxic data from crashed apps.</p><p>to handle untrusted memory data safety.</p><p>without this user every crash would be handled by root.</p><p>it&#39;s used by kernal, dev/admins</p><p>key=CAP_SYS_PTRACE(inspect memory.)</p><h3><strong>8. systemd-oom -&gt; emergency medic</strong></h3><p>user job is to monitor the system and kill off greedy processes before the entire computer freezes or crashes due to lack of RAM.</p><p>Resource Monitoring for RAM and SWAP space.</p><p>used by Cgroups.</p><p>CAP_KILL=terminates other process</p><h3><strong>9. systemd-journal-remote</strong> (sometimes)</h3><h4>👉 These users exist because <strong>services drop privileges</strong> for security.</h4><h3># Core daemons (tightly coupled) of root user.</h3><p>These are managed by systemd and form the OS backbone:</p><h3>Boot &amp; lifecycle</h3><h4>1. <strong>systemd</strong>=The manager of all managers. It has every capability.</h4><ul><li>In systemd, everything is a <strong>Unit</strong>. Units are the building blocks that define the system&#39;s &quot;behavior&quot; and &quot;state.&quot;</li><li><strong>Unit File:</strong> Think of this as the <strong>Blueprint</strong>. It is a plain-text configuration file (usually found in /usr/lib/systemd/system</li><li><strong>The Service:</strong> This is a specific type of Unit (ending in .service). It represents a <strong>living process</strong> (a daemon).</li></ul><h4>2. systemd-sysusers</h4><h4>3. systemd-tmpfiles</h4><p>Creates the volatile folders in /run and /tmp that services need to store their sockets and &quot;state&quot; files.</p><h3># Logging/Recording Memory</h3><ul><li>systemd-journald=Records the system’s “Memory” (Logs).</li><li>key=CAP_SYSLOG</li><li>started by root account user.</li></ul><h3># Device management</h3><ul><li>systemd-udevd=Translates raw hardware into /dev/ files.</li><li>key= Hardware Access.</li><li>started by root account user.</li></ul><h3>#Login/session</h3><ul><li>systemd-logind=Manages the “Human” connection to the machine.</li><li>key= CAP_SYS_ADMIN</li><li>started by root account user.</li></ul><h3># IPC</h3><ul><li>dbus-broker or classic dbus-daemon</li><li>The “Nervous System” for inter-app talk.</li><li>key=Socket Ownership</li></ul><h3># Time sync</h3><ul><li>systemd-timesyncd</li></ul><h3># Networking (one of these is usually active)</h3><ul><li>systemd-networkd started by systemd-network user.</li><li>key=CAP_NET_ADMIN, CAP_NET_BIND_SERVICE.</li><li><strong>Used by:</strong> Servers, IoT devices, and “Container Hosts.”</li><li><strong>Why:</strong> Unlike the heavy “NetworkManager” used on desktops, networkd is a <strong>Core Daemon</strong> designed for &quot;Set and Forget&quot; networking. It manages IPs, routes, and bridges with zero bloat.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e51ba1fa74a5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Abstraction Layer or Service Mapper,]]></title>
            <link>https://medium.com/@aravind.mutthe/operating-systems-and-system-architecture-switches-are-logical-layers-or-service-mappers-that-a7b9bdcfce77?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/a7b9bdcfce77</guid>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Wed, 15 Apr 2026 09:42:41 GMT</pubDate>
            <atom:updated>2026-04-19T18:30:45.726Z</atom:updated>
            <content:encoded><![CDATA[<h3>operating systems and system architecture, “switches” are logical layers or service mappers that abstract underlying data sources from the applications using thema <strong>switch</strong> acts like a “universal remote” for your computer’s identity and configuration data.</h3><p><strong>Service Mappers (Logical Switches)-&gt;Where they live:</strong> <strong>User Space</strong>.</p><p>It is a logical layer that sits between an application (which just wants an answer) and various data sources (where the answer is actually stored).</p><p>The library (like glibc) acts as the switch, looks at the config file, and decides whether to fetch that data from a local file or an LDAP server.</p><h3><strong>1.Service Mapping Switches</strong></h3><p>These switches act as a central “directory” to decide which backend service should handle a specific request.</p><ul><li><strong>NSS (Name Service Switch)</strong>: The primary switch in Unix-like systems. It determines how the system resolves identity information (like users and passwords) and hostnames, mapping these requests to files, DNS, or LDAP.</li><li><strong>PAM (Pluggable Authentication Modules)</strong>: A “switch” for authentication logic. It allows an administrator to change how users are authenticated (e.g., switching from a local password file to a biometric reader or remote server) without modifying the applications.</li><li><strong>VFS (Virtual File System) Switch</strong>: A kernel-level layer that allows the operating system to interact with different file systems (like NTFS, EXT4, or Network File Systems) using a single, unified interface.</li><li><strong>Systemd/Init Service Switch</strong>: Controls the operational state of the machine. It “switches” between different <strong>Targets</strong> or <strong>Run levels</strong>, such as switching from a minimal command-line mode to a full graphical desktop</li></ul><h3><strong>2. Logic &amp; Protocol Switches</strong></h3><p>These mappers handle how data or communication is routed between high-level applications and low-level protocols.</p><ul><li><strong>Socket Switch (WSA / BSD Sockets)</strong>: Maps high-level networking calls to specific transport protocols (like TCP or UDP).</li><li><strong>WFP (Windows Filtering Platform)</strong>: A logic switch for network traffic, allowing security software to “switch” in and out filtering logic for different types of packets.</li><li><strong>NVSwitch / PCIe Switch</strong>: Hardware-level logic switches that map high-speed data paths between GPUs or NVMe drives, ensuring the CPU can communicate with multiple specialized devices simultaneously.</li></ul><h3><strong>3. Firmware &amp; Hardware Initialization Switches</strong></h3><p>These are the most fundamental “switches” that determine the hardware’s operating logic before the OS even loads.</p><ul><li><strong>UEFI / Legacy BIOS Switch</strong>: A firmware-level choice that determines the boot path and hardware initialization method.</li><li><strong>CSM (Compatibility Support Module)</strong>: An emulation switch that allows modern UEFI hardware to “switch” into a mode that supports older, non-UEFI operating systems.</li><li><strong>Secure Boot Switch</strong>: A logic gate that determines if the system will only initialize hardware and software that carries a trusted digital signature.</li></ul><p><strong>The Full Stack Perspective</strong></p><ul><li>If you look at all three of your lists together, you can see the “Path of a Request”:</li></ul><ol><li><strong>Firmware Switch</strong> defines the hardware environment.</li><li><strong>Systemd Switch</strong> starts the necessary background services.</li><li><strong>VFS/Socket Switches</strong> provide the pipes (files/network) for those services to talk.</li><li><strong>NSS/PAM Switches</strong> provide the “Who” and “How” for security.</li></ol><p>To “turn on or off” these switches, you aren’t flipping a physical switch; you are <strong>reconfiguring the logic</strong> the system uses to find data or verify users.</p><h3>The Brief Explanation</h3><p>In a computer, a “switch” is a <strong>decision-maker</strong>. To “turn it on or off,” you modify its <strong>ruleset</strong>.</p><ul><li><strong>For Libraries (NSS/PAM):</strong> You edit a text file to tell the library which “plug-in” to use. If you remove “ldap” from the file, you’ve effectively “turned off” the LDAP switch.</li><li><strong>For the Kernel (VFS):</strong> You use commands like mount or umount to tell the OS to include or ignore a specific storage path.</li><li><strong>For Services (Systemd):</strong> You use the systemctl command to tell the &quot;Master Switch&quot; (PID 1) to change the system&#39;s entire operational state.</li></ul><h3>Management Summary Table</h3><p>Switch What it Controls How to “Turn On/Off” Primary Tool/File</p><p><strong>NSS </strong>Where the system looks for <strong>Identity</strong> (Users/Hosts).</p><p>Add or remove keywords (like dns or ldap) from the config list./etc/nsswitch.conf</p><p><strong>PAM </strong>How the system <strong>Authenticates</strong> you (Passwords/MFA).</p><p>Comment out (#) a module line or use a configuration tool./etc/pam.d/ or pam-auth-update</p><p><strong>VFS </strong>Which <strong>File Systems</strong> are accessible. Use mount to turn a drive &quot;on&quot; and umount to turn it &quot;off.&quot;mount / umount / /etc/fstab</p><p><strong>Systemd </strong>The <strong>System State</strong> (CLI vs. GUI).Use the isolate command to jump between different system targets.systemctl isolate [target]</p><p><strong>HardwareDevice Logic</strong> (Secure Boot, TPM).Toggle settings in the firmware menu before the OS starts.</p><p><strong>UEFI/BIOS Menu</strong> (F2/Del)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a7b9bdcfce77" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Linux Virtual File Systems]]></title>
            <link>https://medium.com/@aravind.mutthe/linux-virtual-file-systems-aabd75b767b1?source=rss-19aa7ffc5f3e------2</link>
            <guid isPermaLink="false">https://medium.com/p/aabd75b767b1</guid>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[memory-management]]></category>
            <category><![CDATA[linux-file-system]]></category>
            <dc:creator><![CDATA[aravind.mutthe]]></dc:creator>
            <pubDate>Fri, 27 Jun 2025 22:11:29 GMT</pubDate>
            <atom:updated>2025-06-27T22:21:54.233Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Bm1oHpQC4BV_G2_brrTLyg.jpeg" /></figure><p>In Linux, these directories: <strong>/proc, /sys, /dev</strong></p><ul><li>Are <strong>not real files on disk</strong> — they exist only in memory (RAM).</li><li>Are <strong>managed by the kernel</strong>, not your hard drive.</li><li>Provide <strong>dynamic, real-time views</strong> of the system’s hardware, kernel, and processes.</li><li>Allow you to <strong>read or modify</strong> certain kernel settings without needing special system calls or reboots.</li></ul><h3><strong>/proc — Process and Kernel Information:</strong></h3><p><strong>What is /proc? </strong>A live <strong>status dashboard</strong> for the kernel and processes</p><ul><li><strong>Filesystem: procfs</strong></li><li><strong>Mount point: </strong>/proc</li><li><strong>Purpose:</strong> A <strong>window into the kernel’s and processes’ live state</strong>.</li><li><strong>Created by: </strong>Kernel</li></ul><p><strong>Uses:</strong></p><ul><li>Monitors system health (<strong>top, ps, htop, uptime</strong> all use it).</li><li>/proc shows <strong>processes</strong></li><li>Debugging processes and open files</li><li>Adjusting kernel runtime settings</li><li>📌 Persistent? No, regenerated at boot.</li></ul><h3><strong>/sys — Kernel Devices &amp; Drivers:</strong></h3><p><strong>What is /sys?</strong>A <strong>device manager</strong> with knobs for hardware &amp; drivers</p><ul><li><strong>Filesystem:</strong> <strong>sysfs</strong></li><li><strong>Mount point:</strong> /sys</li><li><strong>Purpose:</strong> Exposes the <strong>kernel’s device model and driver attributes</strong>.</li><li><strong>Created by:</strong> Kernel</li></ul><p><strong>Uses:</strong></p><ul><li>Hardware detection tools, <strong>udev</strong> rules, hotplugging.</li><li>/sys shows <strong>hardware/devices</strong>.</li><li>Runtime driver debugging</li><li>Manual tuning (if root)</li><li>📌 Persistent? No, rebuilt every boot.</li></ul><h3><strong>/dev — Device Nodes &amp; I/O Access:</strong></h3><p><strong>What is /dev? Doors to the hardware itself</strong> — if you open a file in /dev, you’re talking directly to that device</p><ul><li><strong>Filesystem:</strong> <strong>devtmpfs</strong></li><li><strong>Mount point:</strong> /dev</li><li><strong>Purpose:</strong> Provides <strong>special files that act as access points</strong> to hardware devices and some virtual devices.</li><li><strong>Created by:</strong> Kernel &amp; udev (user-space device manager)</li></ul><p><strong>uses:</strong></p><ul><li>Direct I/O to hardware</li><li>Mounting drives: mount /dev/sdb1 /mnt</li><li>Discard output: command &gt; /dev/null</li><li>VPN tunnels and virtual network devices</li><li>📌 Persistent? No, created dynamically at boot or hotplug via udev.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aabd75b767b1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>