<?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[techbeatly - Medium]]></title>
        <description><![CDATA[Articles and how-to’s on #AI #Automation #Containerization and other #devops technologies.   https://www.techbeatly.com - Medium]]></description>
        <link>https://medium.com/techbeatly?source=rss----ff49a217e60---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>techbeatly - Medium</title>
            <link>https://medium.com/techbeatly?source=rss----ff49a217e60---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 19:14:53 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/techbeatly" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[NVIDIA KAI Scheduler]]></title>
            <link>https://medium.com/techbeatly/nvidia-kai-scheduler-b4dec0d0f8f7?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/b4dec0d0f8f7</guid>
            <dc:creator><![CDATA[Praveenpm]]></dc:creator>
            <pubDate>Tue, 10 Nov 2026 11:37:30 GMT</pubDate>
            <atom:updated>2026-11-10T11:37:30.356Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R2mPOWTZq9EPsnhLM8Gs1w.png" /></figure><p>It’s 9 AM on Monday. Your shiny new GPU cluster is fully installed.</p><p>And the complaints have already started.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*45XgyO3ysBLpeVw-9UocyQ.png" /></figure><p>Sound familiar? I’ve seen versions of this on almost every shared GPU cluster I’ve worked on.</p><p>Here’s the uncomfortable truth: <strong>buying more GPUs won’t fix any of it.</strong> These are scheduling problems. And the default Kubernetes scheduler was never built to solve them.</p><p>That’s exactly why NVIDIA built <strong>KAI Scheduler</strong>.</p><p>By the end of this post, you’ll be able to explain KAI to your team, or a customer in under a minute. No PhD in scheduling theory required. Just one analogy.</p><blockquote><strong><em>The 10-second version:</em></strong><em> KAI is NVIDIA’s open-source Kubernetes scheduler for AI. It gives every team a guaranteed share of GPUs, lends idle GPUs to whoever needs them, and starts big training jobs all at once instead of piece by piece.</em></blockquote><h3>First, picture a busy restaurant</h3><p>Before any Kubernetes jargon, imagine the host at a popular restaurant on a Saturday night.</p><p>The host has one job: <strong>keep every table full, without anyone feeling cheated.</strong> Regulars have reservations. Walk-ins want a seat. A birthday party of eight needs to sit together. A solo diner just wants coffee.</p><p>A good host juggles all of this smoothly. <strong>KAI is that host, for your GPUs.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Zuw9sh-Wvha9_Hqqo1ADyA.png" /></figure><p>Got it? Great. Now let’s answer the six questions everyone asks.</p><h3>1. What is NVIDIA KAI Scheduler?</h3><p>Every Kubernetes cluster has a <strong>scheduler</strong>: the component that decides which machine runs each new workload.</p><p>The built-in one is great for web apps. But it treats a $30,000 GPU the same way it treats a CPU core.</p><p><strong>KAI Scheduler is a Kubernetes scheduler built specifically for AI workloads.</strong> It decides which team’s training job, inference service or notebook gets which GPUs, when, and for how long.</p><p>Three quick facts:</p><ul><li><strong>It came from Run:ai.</strong> KAI is the scheduling engine inside NVIDIA Run:ai. NVIDIA open-sourced it in April 2025 under Apache 2.0.</li><li><strong>It belongs to the community.</strong> It’s now a CNCF Sandbox project at github.com/kai-scheduler/KAI-Scheduler.</li><li><strong>It powers open-source Run:ai.</strong> In October 2026, NVIDIA open-sourced the entire Run:ai platform, with KAI at its core.</li></ul><blockquote><em>KAI lets many teams share one AI cluster fairly, and keeps every GPU busy.</em></blockquote><h3>2. Why do we need it?</h3><p>Go back to those four Monday-morning complaints. Each one maps to something the default scheduler simply can’t do.</p><ul><li><strong>It can’t seat a whole party at once.</strong> It places pods one by one, so a 32-worker job can start 20 workers that sit idle, holding GPUs hostage.</li><li><strong>It doesn’t know about reservations.</strong> It has no idea which team owns which GPUs or who’s entitled to what.</li><li><strong>It can’t share a table.</strong> To it, a GPU is a whole number. A notebook asking for a sliver gets the entire device.</li><li><strong>It never lends empty tables.</strong> Idle GPUs stay idle, even while another team’s queue grows.</li></ul><p>The result? <strong>Expensive GPUs sitting idle while jobs wait in line.</strong> That’s the worst of both worlds.</p><h3>3. How does it work?</h3><p>KAI’s architecture fits in three layers: what comes in, how KAI decides, and where work lands.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ta87wym8U_-pwy2tlo7hHg.png" /></figure><h3>Layer 1: What comes in</h3><p>Your favorite tools keep working: PyTorch and Kubeflow jobs, Ray, model serving, Jupyter notebooks, Argo pipelines. Opting in takes just two lines:</p><pre>metadata:<br>  labels:<br>    kai.scheduler/queue: team-llm   # which team&#39;s &quot;reservation&quot;<br>spec:<br>  schedulerName: kai-scheduler      # let the host seat this pod</pre><h3>Layer 2: How KAI decides</h3><p><strong>Podgroups (the party).</strong> KAI bundles related pods into one group. The whole group gets seated together, or it waits. This is called <em>gang scheduling</em>, and the built-in podgrouper sets it up automatically for common frameworks.</p><p><strong>Queues (the reservations).</strong> Each team gets a queue with four simple dials:</p><ul><li><strong>Quota:</strong> GPUs you’re always guaranteed.</li><li><strong>Over-quota weight:</strong> your share of idle GPUs you can borrow.</li><li><strong>Limit:</strong> the ceiling you can never cross.</li><li><strong>Priority:</strong> who gets to borrow first.</li></ul><blockquote><strong><em>Remember it like this:</em></strong><em> quota is what you’re promised, over-quota is what you can borrow, and limit is the ceiling.</em></blockquote><p><strong>The loop (the host’s routine).</strong> KAI constantly takes a snapshot of the cluster, works out each team’s fair share, and then acts. And just like a good host, it always tries the polite move first.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gJyOcsEUsBXR8Zhv-Iwg2A.png" /></figure><p>That ordering matters. KAI will happily <strong>rearrange</strong> pods to make room before it ever <strong>evicts</strong> one.</p><h3>Layer 3: Where work lands</h3><p>The binder places pods on the chosen GPU nodes, which are prepared by the NVIDIA GPU Operator. And the default Kubernetes scheduler? It keeps running for everything else. KAI only seats the guests who ask for it.</p><h3>4. What is it actually used for?</h3><p>Here’s what teams use KAI for day to day:</p><ul><li><strong>Sharing one cluster across many teams</strong>, each with a guaranteed slice.</li><li><strong>Running distributed training safely</strong>, with every worker starting together.</li><li><strong>Splitting a GPU</strong> for notebooks and small models, for example.</li><li><strong>Protecting production inference</strong>, so customer-facing services are never kicked off by another team.</li><li><strong>Keeping multi-node jobs close together</strong> with topology-aware placement for faster communication.</li><li><strong>Staying fair over time.</strong> Time-based fair share remembers who used what last week, so occasional big jobs still get their turn.</li></ul><blockquote><em>⚠️ </em><strong><em>Good to know:</em></strong><em> by default, KAI </em>tracks<em> GPU fractions but doesn’t </em>enforce<em> memory limits between pods on a shared GPU. For hard isolation, use an enforcing mode like HAMi or NvFractions, or NVIDIA’s GPU Fractioning Operator.</em></blockquote><h3>Where should you use it?</h3><p>The simple rule: <strong>if more than one team fights over the same GPUs on Kubernetes, KAI will help.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1N9eVgE_nNToLtmJ0y88sQ.png" /></figure><p>Momentum is building fast, too. Crusoe, Nebius, Mirantis, Rafay and Spectro Cloud are among the providers integrating open-source Run:ai, with KAI at its core.</p><h3>Save this cheat sheet</h3><p>Everything above, on one page. Screenshot it, share it, pin it next to your cluster dashboard.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9969HbR2CoSgit4TMDN6Dg.png" /></figure><h3>References</h3><ul><li><a href="https://developer.nvidia.com/blog/nvidia-open-sources-runai-scheduler-to-foster-community-collaboration/">NVIDIA Technical Blog: NVIDIA Open Sources Run:ai Scheduler to Foster Community Collaboration</a></li><li><a href="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog: NVIDIA Run:ai Is Now Open Source as Part of NVIDIA DSX AI Factory Platform</a></li><li><a href="https://developer.nvidia.com/blog/ensuring-balanced-gpu-allocation-in-kubernetes-clusters-with-time-based-fairshare/">NVIDIA Technical Blog: Ensuring Balanced GPU Allocation with Time-Based Fairshare</a></li><li><a href="https://github.com/kai-scheduler/KAI-Scheduler">KAI Scheduler on GitHub</a></li><li><a href="https://github.com/kai-scheduler/KAI-Scheduler/blob/main/docs/scheduling-deep-dive/README.md">KAI docs: Scheduling Deep Dive</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b4dec0d0f8f7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/nvidia-kai-scheduler-b4dec0d0f8f7">NVIDIA KAI Scheduler</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[NVIDIA KV Cache: The Memory Trick That Makes AI 10× Faster]]></title>
            <link>https://medium.com/techbeatly/nvidia-kv-cache-the-memory-trick-that-makes-ai-10-faster-1f23eefd2a65?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/1f23eefd2a65</guid>
            <dc:creator><![CDATA[Praveenpm]]></dc:creator>
            <pubDate>Sun, 31 May 2026 17:02:01 GMT</pubDate>
            <atom:updated>2026-05-31T17:02:00.903Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1008/1*UAhmJh2mHZCP8VgCmcZlJQ.png" /></figure><p><em>One optimization. Massive impact. Here’s what every AI builder needs to know.</em></p><p><strong>For:</strong> Solution Architect — AI · MLOps · LLMOps · AIOps · Platform Engineering.</p><h3>The Problem</h3><p>Your LLM is secretly doing something absurd.</p><p>Every time it generates a single word, it <strong>re-reads your entire conversation from the beginning.</strong> Word by word. Every turn. Every time.</p><p>Imagine asking your assistant a question — and they re-read every email in their inbox before replying. That’s what’s happening inside your AI without KV Cache.</p><p>The result? Latency spikes. GPU bills that make your CFO cry. And AI products that feel sluggish just when users need them most.</p><p><strong>KV Cache fixes this entirely.</strong></p><h3>What Is KV Cache?</h3><p>Inside every transformer model, three vectors are computed for each token:</p><p><strong>Q — Query</strong> → “What am I looking for?”</p><p><strong>K — Key</strong> → “What do I contain?”</p><p><strong>V — Value</strong> → “What’s my actual content?”</p><p>Without KV Cache → K and V are <strong>recomputed from scratch</strong> at every generation step.</p><p>With KV Cache → K and V are <strong>computed once, stored, and reused.</strong></p><blockquote><em>Think of it like RAM for your AI’s short-term memory. Compute once. Read many times.</em></blockquote><h3>How the Data Actually Flows</h3><p><strong>STEP 1 — PREFILL</strong> <em>(runs once, on your full prompt)</em></p><pre>👤 &quot;Summarize this contract and flag risks&quot;<br>            │<br>            ▼<br>    [ Tokenizer ]<br>    [T1][T2][T3]···[TN]<br>            │<br>            ▼<br>┌─────────────────────────┐<br>│   Transformer Layers    │<br>│   Compute Q + K + V     │  ← happens for ALL tokens<br>│   for every token       │<br>└────────┬────────┬───────┘<br>         │        │<br>   ┌─────▼──┐ ┌───▼─────┐<br>   │ K Cache│ │ V Cache │  ← saved to GPU memory (HBM)<br>   │ 🗄 ████│ │ 🗄 ████ │<br>   └────────┘ └─────────┘<br>         ▲        ▲<br>         └──┐  ┌──┘<br>   PagedAttention manages<br>   these like OS page tables<br>   (no fragmentation, no waste)</pre><p><strong>STEP 2 — DECODE</strong> <em>(loops for every new token)</em></p><pre>New token arrives →  only Q is computed (1 token!)<br>            │<br>            ▼<br>┌──────────────────────────────────┐<br>│  Attention: Q × K_cached         │  ← READ from cache<br>│  → attention scores              │     no recompute!<br>│  scores × V_cached               │<br>│  → context vector                │<br>└──────────────┬───────────────────┘<br>               │<br>               ▼<br>      [ FFN + Projection ]<br>               │<br>               ▼<br>       ✅  Next token out</pre><pre>       New K, V appended ──→  K Cache / V Cache grows by 1<br>               │<br>               └──────────────────────────────────┐<br>                        loop ↺ until done          │<br>               ┌────────────────────────────────── ┘<br>               ▼<br>       ✅  Next token out  ...  ✅  ...  ✅  Done!</pre><p><strong>The bottom line:</strong> Prefill runs <strong>once.</strong> Decode reads from cache. Your GPU stops doing the same work over and over.</p><h3>Why This Changes Everything for Enterprise AI</h3><p>Here’s what happens to your numbers when you turn KV Cache on:</p><p><strong>Time to First Token:</strong> 3–5 sec → 200–400 ms | <strong>10× faster</strong></p><p><strong>Concurrent Users / GPU:</strong> 50 → 400–500 | <strong>8–10× more</strong></p><p><strong>GPU Cost per Query:</strong> $0.10 → $0.01–0.02 | <strong>80% cheaper</strong></p><p><strong>Max Usable Context:</strong> ~4K tokens → 128K+ tokens | <strong>32× longer</strong></p><p><strong>P99 Latency:</strong> 10–15 sec → 1–2 sec | <strong>8× faster</strong></p><p>One optimization. Those numbers. No new hardware.</p><h3>Real-World Impact: The HR Chatbot Example</h3><p>A global enterprise deploys an HR assistant for <strong>50,000 employees.</strong> Every conversation starts with a 2,000-token company policy context.</p><p><strong>Without KV Cache:</strong></p><pre>50,000 users × full prefill each = 50,000 GPU-heavy computations</pre><p><strong>With Prefix Caching:</strong></p><pre>User #1       → prefill runs  ✓  KV stored<br>User #2       → cache hit     ✓  instant<br>User #3       → cache hit     ✓  instant<br>  ·<br>  ·<br>User #50,000  → cache hit     ✓  instant</pre><p><strong>Result:</strong> 1 compute. 49,999 free rides. Cost drops ~95%.</p><h3>Where to Apply It in Your Stack</h3><p><strong>Agentic AI Pipelines</strong></p><p>Multiple agents (planner → researcher → coder → reviewer) share one system prompt. Cache it once. All agents run faster. Multi-agent cost drops.</p><p><strong>RAG Systems</strong></p><p>Your top-100 retrieved documents appear in thousands of queries daily. Pre-warm the KV Cache at startup. Every RAG query becomes a near-instant cache hit.</p><p><strong>AIOps &amp; Log Analysis</strong></p><p>Load a 128K-token incident log once. Your SRE team asks 20 different questions -all answered from cache in under a second. No re-encoding 50MB of logs per question.</p><p><strong>Long Multi-Turn Chat</strong></p><p>100-turn customer conversations. Each new message only computes the latest turn. The full history lives in cache. Coherent, fast, cheap.</p><h3>The NVIDIA Stack</h3><pre>┌──────────────────────────────────────────────────┐<br>│              NVIDIA KV Cache Stack               │<br>├──────────────┬───────────────────────────────────┤<br>│ vLLM         │ Prefix caching + PagedAttention   │<br>│ TensorRT-LLM │ FP8 KV quantization (2× memory)  │<br>│ Dynamo       │ Disaggregated prefill/decode       │<br>│ Triton       │ Cache hit rate metrics + serving  │<br>│ H200 / GH200 │ 480GB+ HBM pool for huge contexts │<br>└──────────────┴───────────────────────────────────┘</pre><p><strong>Start here:</strong> vLLM with enable_prefix_caching=True — one line, immediate gains, zero infrastructure change.</p><h3>The One Metric to Rule Them All</h3><p>In your LLMOps dashboard, track this above everything else:</p><pre>Prefix Cache Hit Rate  →  target: &gt; 80%</pre><pre>  &lt; 50%  ⚠️  Something&#39;s wrong. Prompts too varied or too dynamic.<br>  50–79% 🟡  Room to optimize. Standardize system prompts.<br>  80–95% ✅  Healthy. You&#39;re saving serious compute.<br>  &gt; 95%  🚀  Excellent. Near-zero prefill cost at scale.</pre><p>If your hit rate is low — standardize your system prompts, chunk your RAG documents consistently, and consider SGLang’s RadixAttention for tree-structured prefix sharing.</p><h3>Conclusion</h3><p>KV Cache is not a nice-to-have. For any AI system running in production at scale, <strong>it is the foundation.</strong></p><p>The organizations winning with AI in 2025 are not necessarily running bigger models — they are running smarter infrastructure. KV Cache is the clearest example of that.</p><p><strong>Three things to do this week:</strong></p><ol><li>Enable prefix_caching=True in your vLLM deployment</li><li>Add prefix_cache_hit_rate to your Grafana dashboard</li><li>Standardize every system prompt across your agent fleet</li></ol><p>Your GPU bill will thank you. Your users will feel it.</p><p>#AI #MachineLearning #LLMOps #MLOps #NvidiaAI #GenerativeAI #AIOps #EnterpriseAI #SolutionArchitecture #AIInfrastructure #KVCache #LLM #GPUComputing</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f23eefd2a65" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/nvidia-kv-cache-the-memory-trick-that-makes-ai-10-faster-1f23eefd2a65">NVIDIA KV Cache: The Memory Trick That Makes AI 10× Faster</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[NVLink]]></title>
            <link>https://medium.com/techbeatly/nvlink-1ea2247674d3?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/1ea2247674d3</guid>
            <dc:creator><![CDATA[Praveenpm]]></dc:creator>
            <pubDate>Sun, 24 May 2026 16:31:26 GMT</pubDate>
            <atom:updated>2026-05-24T16:31:26.225Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JjiRIIx1xKet_T4AtvZAhA.png" /><figcaption>NVIDIA NVLink: Why Your GPUs Need to Talk Faster</figcaption></figure><h3>The hidden bottleneck that determines whether your AI system scales — or stalls</h3><p><em>Published for: MLOPs, LLMOps, AI Engineers, Solution A</em>rchitectures, <em>AI Infrastructure</em></p><p>Imagine you hired 72 of the world’s fastest workers and put them in the same building. They’re brilliant. They work fast. But there’s one problem — they can only communicate by shouting across the hallway.</p><p>That’s what multi-GPU AI systems looked like before NVLink.</p><p>GPUs are powerful. But when multiple GPUs need to work together on a large AI model, they constantly need to share data with each other. The speed of that communication determines whether the system scales — or whether all that expensive compute sits idle, waiting.</p><p><strong>NVLink is NVIDIA’s solution to that communication problem.</strong></p><p>In this post, I’ll walk you through what NVLink is, why it matters, how it evolved, and what it means for anyone building or running AI systems today.</p><h3>The Problem: GPUs Are Powerful but Isolated</h3><p>Every GPU has its own memory — called HBM (High Bandwidth Memory). When you’re training a large language model across 8 GPUs, those GPUs need to constantly share:</p><ul><li>Gradients (what the model learned in each training step)</li><li>Activations (intermediate results during forward pass)</li><li>Model weight shards (pieces of a model too large to fit in one GPU)</li></ul><p>Before NVLink, all of this data moved over <strong>PCIe</strong> — the standard interface used for connecting components inside a server.</p><p>PCIe Gen 4 delivers about <strong>32 GB/s</strong>. That sounds fast. But when you’re running a 70B-parameter model, a single inference request can generate <strong>20 GB of synchronization data</strong> that must be exchanged between every GPU — for every request, every layer, every step.</p><p>At that scale, PCIe becomes the bottleneck. GPUs go idle waiting for data. The system underperforms. More GPUs make it worse, not better.</p><h3>The Solution: NVLink</h3><p><strong>NVLink is NVIDIA’s high-speed, direct GPU-to-GPU interconnect.</strong></p><p>Instead of sending data through the server’s PCIe bus and bouncing it through the CPU, NVLink creates a <strong>direct, dedicated highway</strong> between GPUs. Data moves faster, with lower latency, and without interrupting the CPU.</p><p>Three things make NVLink fundamentally different from PCIe:</p><p><strong>1. Massive bandwidth</strong> NVLink 5 (the current generation in Blackwell GPUs) delivers <strong>1,800 GB/s per GPU</strong> — that’s 28× more bandwidth than PCIe Gen 4, and 14× more than PCIe Gen 5.</p><p><strong>2. Peer memory access</strong> With NVLink, GPU A can read and write directly into GPU B’s memory — no CPU involved, no data copying overhead. It’s as if both GPUs share the same memory space.</p><p><strong>3. Works automatically with CUDA</strong> You don’t need to rewrite your code. If you have NVLink-connected GPUs and call cudaDeviceEnablePeerAccess(), CUDA automatically routes data transfers over NVLink instead of PCIe. Your existing PyTorch, JAX, or TensorFlow workloads benefit without changes.</p><h3>A Simple Analogy</h3><p>Think of PCIe as a city road network — shared, congested, and managed by traffic lights (the CPU).</p><p>NVLink is a private, direct expressway between two buildings. No traffic lights. No shared lanes. Full speed, both directions, at all times.</p><p>When you’re moving millions of cars per hour (gradient tensors), the difference between a shared road and a private expressway is the difference between a working system and a gridlocked one.</p><h3>NVSwitch: The Traffic Exchange for Many GPUs</h3><p>NVLink is great for connecting two GPUs. But what about 8? Or 72?</p><p>If you connect 8 GPUs with point-to-point NVLink cables, the bandwidth gets shared when multiple pairs need to talk simultaneously. GPU A’s conversation with GPU B eats into GPU C’s available bandwidth to GPU D.</p><p><strong>NVSwitch solves this.</strong></p><p>NVSwitch is a dedicated switching chip that sits between the GPUs and acts like a <strong>non-blocking crossbar</strong> — every GPU can talk to every other GPU at full NVLink speed, simultaneously, without any of them competing for bandwidth.</p><pre>Without NVSwitch: Adding more GPUs splits available bandwidth.<br>With NVSwitch: Every GPU always gets its full NVLink bandwidth, regardless of how many others are communicating at the same time.</pre><p>In an 8-GPU HGX H100 server, four NVSwitch chips provide this full all-to-all connectivity. Each NVSwitch chip has a total bidirectional bandwidth of 25.6 terabits per second.</p><p>That’s not a typo.</p><h3>Six Generations in One Table</h3><p>NVLink has evolved dramatically since its introduction in 2016:</p><pre>| Generation | Architecture     | Year  | Bandwidth per GPU |<br>| ---------- | ---------------- | ----- | ----------------- |<br>| NVLink 1.0 | Pascal (P100)    | 2016  | 80 GB/s           |<br>| NVLink 2.0 | Volta (V100)     | 2017  | 300 GB/s          |<br>| NVLink 3.0 | Ampere (A100)    | 2020  | 600 GB/s          |<br>| NVLink 4.0 | Hopper (H100)    | 2022  | 900 GB/s          |<br>| NVLink 5.0 | Blackwell (B200) | 2024  | 1,800 GB/s        |<br>| NVLink 6.0 | Rubin            | 2025+ | 3,600 GB/s        |</pre><p>In 10 years, per-GPU bandwidth has grown <strong>45×</strong>. The trajectory is not slowing down — it’s accelerating alongside the growth of AI model sizes.</p><h3>Why This Matters for AI Workloads</h3><h3>Training Large Models</h3><p>When training a model like GPT-4 or Llama 3 across hundreds of GPUs, each training step requires an <strong>All-Reduce operation</strong> — every GPU sends its gradients to every other GPU and receives the averaged result.</p><p>On a PCIe-based system, this collective communication step becomes the bottleneck. On an NVLink system with NVSwitch, the all-reduce completes at near-theoretical peak bandwidth, and GPU compute cores stay busy instead of waiting.</p><p>NVIDIA’s own measurements show that an 8-GPU all-reduce for a 20 GB tensor completes in ~22ms with NVSwitch — versus ~150ms without. That’s a <strong>7× speedup on a single operation</strong> that repeats thousands of times per training run.</p><h3>Inference on Large Models</h3><p>For serving large models in production — say, a 70B or 405B parameter model — the model must be split across multiple GPUs using <strong>tensor parallelism</strong>. Each GPU holds a slice of the model’s weight matrices.</p><p>During inference, after each layer, every GPU must exchange its partial results with all others before proceeding to the next layer. This happens for every single token generated, for every user request.</p><p>NVLink’s bandwidth is the direct determinant of how fast tokens appear in the user’s chat window. Faster interconnect → faster tokens → better user experience.</p><h3>Long-Context and Agentic AI</h3><p>Modern AI workloads involving very long contexts (128K–1M tokens) or multi-step agentic reasoning generate <strong>KV caches</strong> (key-value caches that store past attention states) that can be too large to fit in a single GPU’s memory.</p><p>NVLink enables KV caches to be distributed across the GPUs in an NVLink domain and accessed at memory speeds — making these workloads tractable on hardware that would otherwise be insufficient.</p><h3>The GB200 NVL72: What NVLink Looks Like at Scale</h3><p>NVIDIA’s current flagship system — the <strong>GB200 NVL72</strong> — is the most powerful NVLink deployment in production today.</p><p>Here’s what it contains:</p><ul><li><strong>72 Blackwell B200 GPUs</strong>, each with 1,800 GB/s NVLink 5 bandwidth</li><li><strong>36 Grace CPUs</strong>, connected to their paired GPUs via NVLink-C2C (a chip-to-chip variant)</li><li><strong>9 NVLink Switch trays</strong> providing the all-to-all fabric</li><li><strong>130 TB/s</strong> total aggregate bandwidth across the rack</li><li>Liquid cooling (required — the thermal density rules out air cooling)</li></ul><p>From the software’s perspective, the entire rack looks like a <strong>single unified accelerator</strong> — one giant compute pool with shared memory. vLLM, TensorRT-LLM, and other inference frameworks treat the 72 GPUs as one coherent system.</p><p>Scaled further, 8 of these racks form a <strong>DGX SuperPOD</strong> — 576 Blackwell GPUs connected in a single NVLink domain with 1 petabyte per second of aggregate bandwidth.</p><h3>NVLink vs. PCIe vs. InfiniBand</h3><p>People often ask: “If I have InfiniBand, do I still need NVLink?”</p><p>Yes. They serve completely different purposes.</p><p>Technology Where It Works What It’s For <strong>NVLink</strong> Inside a rack GPU↔GPU fast communication (tensor shards, gradients) <strong>PCIe</strong> Inside a server GPU↔CPU, GPU↔storage, GPU↔network card <strong>InfiniBand / Ethernet</strong> Across racks and data centers Node↔node communication for large clusters</p><pre>| Technology            | Purpose                 |<br>| --------------------- | ----------------------- |<br>| NVLink                | GPU↔GPU communication   |<br>| PCIe                  | GPU↔CPU / storage       |<br>| InfiniBand / Ethernet | Rack-to-rack networking |</pre><p>Every well-designed AI cluster uses all three. NVLink handles the intense, high-frequency communication <em>within</em> the rack. InfiniBand handles the communication <em>between</em> racks. PCIe handles everything connecting to the host.</p><p>Choosing between them isn’t the right question. Knowing which layer each one serves is.</p><h3>What’s Next: NVLink 6 and NVLink Fusion</h3><h3>NVLink 6 (Rubin Architecture)</h3><p>The next generation, shipping with NVIDIA’s Rubin platform, doubles NVLink 5’s bandwidth to <strong>3,600 GB/s per GPU</strong> — 14× the bandwidth of PCIe Gen 6. Seventy-two Rubin GPUs connected in a Vera Rubin NVL72 rack will have <strong>260 TB/s</strong> of aggregate bandwidth.</p><p>NVLink 6 also introduces in-switch computation via the <strong>SHARP</strong> protocol, where AllReduce operations are partially computed inside the NVLink Switch itself — eliminating additional round-trips to GPU memory.</p><h3>NVLink Fusion</h3><p>Perhaps more strategically significant: in 2025, NVIDIA opened NVLink to third-party silicon via <strong>NVLink Fusion</strong>. AWS has already integrated their Trainium4 custom chip into the NVLink fabric using this technology.</p><p>This means NVLink is evolving from an NVIDIA-exclusive technology into a <strong>standard scale-up interconnect</strong> that heterogeneous AI hardware can plug into. It’s a significant shift for anyone thinking about long-term AI infrastructure strategy.</p><h3>Key Takeaways</h3><p>Here’s what to remember from this post:</p><ul><li><strong>NVLink is a direct GPU-to-GPU interconnect</strong> that replaces PCIe for intra-rack GPU communication, delivering up to 1,800 GB/s per GPU in the current Blackwell generation</li><li><strong>NVSwitch adds all-to-all connectivity</strong> so every GPU in a server or rack communicates at full NVLink speed simultaneously — no sharing, no contention</li><li><strong>It works transparently with CUDA</strong> — frameworks like PyTorch and vLLM use it automatically via NCCL without code changes</li><li><strong>NVLink, PCIe, and InfiniBand operate at different layers</strong> — they complement each other, not compete</li><li><strong>The bandwidth doubles roughly every two years</strong> — NVLink 6 hits 3,600 GB/s per GPU with Rubin in 2025–2026</li><li><strong>NVLink Fusion opens the technology to third-party chips</strong>, signaling its evolution into an industry-standard scale-up fabric</li></ul><p><strong>Tags:</strong> NVIDIA · GPU · AI Infrastructure · Machine Learning · Deep Learning · MLOps · LLMOps · Platform Engineering</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ea2247674d3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/nvlink-1ea2247674d3">NVLink</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Inside NVIDIA Blackwell: The GPU Architecture Redefining the Age of AI]]></title>
            <link>https://medium.com/techbeatly/inside-nvidia-blackwell-the-gpu-architecture-redefining-the-age-of-ai-5892679ae82a?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/5892679ae82a</guid>
            <dc:creator><![CDATA[Praveenpm]]></dc:creator>
            <pubDate>Sat, 16 May 2026 06:39:08 GMT</pubDate>
            <atom:updated>2026-05-26T08:52:38.128Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IGYm3KCBR2Gjzn8s8TTIOQ.png" /></figure><p><em>How a two-die chiplet, 208 billion transistors, and a rack-scale supercomputer are reshaping every layer of the AI/ML stack — from training trillion-parameter models to real-time LLM inference.</em></p><h3>The Shift Nobody Saw Coming</h3><p>For more than a decade, GPU evolution followed a familiar pattern: build larger chips, add more CUDA cores, and push higher FLOPS. That strategy worked remarkably well — until physics imposed a hard limit.</p><p>Modern photolithography tools can expose only about <strong>800 mm²</strong> of silicon in a single reticle shot. Beyond that, manufacturing a larger monolithic die becomes impractical.</p><p>Rather than fighting the reticle limit, NVIDIA completely redefined what a “single GPU” could be.</p><p>Introduced in 2024, Blackwell is the company’s most significant architectural leap since the introduction of Tensor Cores in the Volta generation. It is not merely a performance upgrade — it is a structural redesign of the GPU itself.</p><p>For engineers working in AI infrastructure, platform engineering, MLOps, LLMOps, and high-performance AI systems, Blackwell introduces an entirely new architectural mindset.</p><h3>1. The Dual-Die Chiplet Architecture</h3><p>The defining innovation of Blackwell is straightforward yet revolutionary:</p><p><strong>Blackwell is not a single die GPU.</strong></p><p>Instead, it combines <strong>two reticle-limited dies</strong>, each containing approximately <strong>104 billion transistors</strong>, connected through an ultra-fast <strong>10 TB/s chip-to-chip interconnect</strong>.</p><pre>┌──────────────────────────────────────────┐<br>│             B200 GPU Package             │<br>│  ┌──────────┐  10 TB/s  ┌──────────┐    │<br>│  │  Die 1   │◄─────────►│  Die 2   │    │<br>│  │  104B tx │  C2C Link │  104B tx │    │<br>│  └──────────┘           └──────────┘    │<br>│         208B transistors total           │<br>│         TSMC 4NP process node            │<br>└──────────────────────────────────────────┘</pre><p>To developers using CUDA, PyTorch, TensorRT-LLM, or vLLM, the hardware behaves like a single GPU. NVIDIA achieves this through synchronized memory management units, hardware memory barriers, and high-speed die communication layers that abstract the underlying complexity.</p><p>This architectural decision enables a massive jump from Hopper’s 80 billion transistors to Blackwell’s 208 billion — without violating manufacturing constraints.</p><h3>2. Compute Hierarchy: 148 Streaming Multiprocessors</h3><p>Blackwell organizes compute resources into a scalable hierarchy:</p><pre>B200 GPU<br> └── 8 Graphics Processing Clusters (GPCs)<br>       └── Each GPC → 8 Texture Processing Clusters (TPCs)<br>             └── Each TPC → 2 Streaming Multiprocessors (SMs)<br>                   └── Each SM contains:<br>                         ├── 128 CUDA Cores<br>                         ├── 4 × 5th-Gen Tensor Cores<br>                         ├── Shared Memory / L1 Cache<br>                         └── Tensor Memory (TMEM)</pre><p>The architecture includes <strong>148 Streaming Multiprocessors</strong>, compared to Hopper’s 132 SMs.</p><p>However, the real performance leap is not the SM count itself — it is the redesign of the Tensor Core pipeline and memory hierarchy inside each SM.</p><h3>3. Fifth-Generation Tensor Cores and Native FP4</h3><p>Blackwell introduces <strong>native FP4 and FP6 precision formats</strong> through its fifth-generation Tensor Cores.</p><p>PrecisionHopper H100Blackwell B200FP3267 TFLOPS~150 TFLOPSTF32989 TFLOPS~2,250 TFLOPSFP161,979 TFLOPS~4,500 TFLOPSFP83,958 TFLOPS~9,000 TFLOPSFP4Not Supported~18,000 TFLOPS</p><pre>| Precision | Hopper H100  | Blackwell B200 | Gain                      |<br>|-----------|--------------|----------------|------                     |<br>| FP64      | 34 TFLOPS    | ~30 TFLOPS     | — (intentionally reduced) |<br>| FP32      | 67 TFLOPS    | ~150 TFLOPS    | ~2.2×                     |<br>| TF32      | 989 TFLOPS   | ~2,250 TFLOPS  | ~2.3×                     |<br>| FP16      | 1,979 TFLOPS | ~4,500 TFLOPS  | ~2.3×                     |<br>| FP8       | 3,958 TFLOPS | ~9,000 TFLOPS  | ~2.3×                     |<br>| **FP6**   | ❌ None      | ~12,000 TFLOPS | ∞                          |<br>| **FP4**   | ❌ None      | ~18,000 TFLOPS | ∞                          |</pre><p>The strategic focus is clear: Blackwell is optimized specifically for AI workloads rather than traditional FP64-heavy scientific HPC workloads.</p><p>The significance of FP4 is transformative:</p><ul><li>Models require dramatically less memory bandwidth</li><li>More parameters fit into GPU memory</li><li>Inference throughput increases substantially</li><li>Power efficiency improves significantly</li></ul><p>For large language models such as Llama-3 70B, FP4 enables single-GPU serving scenarios that previously required multi-GPU deployments.</p><h3>4. Transformer Engine 2.0</h3><p>Blackwell also introduces the next evolution of the Transformer Engine.</p><p>The original Hopper Transformer Engine focused on FP8 acceleration. Blackwell expands this capability through <strong>micro-tensor scaling</strong> using MXFP4 and MXFP6 formats.</p><pre>Model Weights<br>      │<br>      ▼<br>Micro-Tensor Scaling<br>      │<br>      ▼<br>FP4 / FP6 Tensor Core Operations<br>      │<br>      ▼<br>BF16 / FP32 Accumulation</pre><p>Instead of assigning a single scaling factor per tensor, Blackwell dynamically scales smaller tensor blocks independently. This preserves numerical stability while aggressively reducing precision.</p><p>The outcome is highly significant for production AI systems:</p><ul><li>FP4 inference achieves &gt;99.5% model accuracy retention</li><li>Quantization becomes production-ready</li><li>Lower latency and reduced memory usage become standard deployment patterns</li></ul><p>This changes quantization from an experimental optimization into a mainstream deployment strategy.</p><h3>5. Tensor Memory (TMEM): A New Memory Layer</h3><p>One of the most underrated innovations in Blackwell is <strong>Tensor Memory (TMEM)</strong>.</p><p>TMEM introduces a dedicated on-chip memory layer optimized specifically for tensor operations and transformer workloads.</p><h3>Blackwell Memory Hierarchy</h3><pre>Registers<br>Shared Memory / L1<br>TMEM (NEW)<br>L2 Cache<br>HBM3e<br>NVLink</pre><p>TMEM enables transformer attention data and quantized weights to remain closer to the compute pipeline, minimizing expensive HBM memory accesses.</p><p>For LLM inference workloads, where KV-cache movement dominates latency, TMEM provides substantial performance and energy-efficiency improvements.</p><h3>6. HBM3e: 192 GB and 8 TB/s Bandwidth</h3><p>Memory bandwidth has long been the bottleneck for large-scale AI inference.</p><p>Blackwell directly addresses this challenge with:</p><ul><li><strong>192 GB HBM3e memory</strong></li><li><strong>8 TB/s memory bandwidth</strong></li></ul><p>GPUMemoryBandwidthA10080 GB2.0 TB/sH10080 GB3.35 TB/sH200141 GB4.8 TB/sB200192 GB8.0 TB/s</p><p>This enables:</p><ul><li>Larger models per GPU</li><li>Longer context windows</li><li>Higher token generation throughput</li><li>Reduced multi-GPU dependency</li></ul><p>For modern generative AI systems, bandwidth is often more important than raw FLOPS. Blackwell significantly improves both.</p><h3>7. NVLink 5 and the GB200 NVL72 Rack</h3><p>Blackwell extends beyond individual GPUs into rack-scale architecture.</p><p>NVLink 5 delivers:</p><ul><li><strong>1.8 TB/s bidirectional bandwidth per GPU</strong></li><li>2× the bandwidth of Hopper NVLink 4</li></ul><p>The flagship deployment model is the <strong>GB200 NVL72</strong>, a rack containing:</p><ul><li>36 Grace CPUs</li><li>72 Blackwell GPUs</li><li>~13.8 TB total HBM memory</li><li>130 TB/s aggregate NVLink bandwidth</li></ul><pre>GB200 NVL72<br> ├── 36 Grace CPUs<br> ├── 72 B200 GPUs<br> ├── 130 TB/s NVLink Fabric<br> └── 1.44 ExaFLOPS FP4 AI Compute</pre><p>From frameworks such as PyTorch or NeMo, the entire rack behaves like a unified AI supercomputer.</p><p>This dramatically reduces communication overhead for trillion-parameter model training and inference.</p><h3>8. Impact on AI Infrastructure and MLOps</h3><p>Blackwell fundamentally changes how AI infrastructure should be designed.</p><h3>Model Serving</h3><ul><li>FP4 becomes the preferred inference precision</li><li>Single-GPU serving becomes viable for larger models</li><li>Long-context serving becomes practical without aggressive KV compression</li></ul><h3>Training</h3><ul><li>Higher throughput per watt</li><li>Faster distributed training</li><li>Reduced networking bottlenecks inside NVLink domains</li></ul><h3>MLOps and Platform Engineering</h3><ul><li>Improved observability through RAS integration</li><li>Faster data pipeline decompression</li><li>Better infrastructure efficiency for multi-tenant AI platforms</li></ul><h3>Cost Efficiency</h3><p>Although Blackwell hardware is more expensive per GPU, its effective <strong>cost-per-token</strong> is significantly lower due to much higher inference throughput and efficiency.</p><h3>The Road Beyond Blackwell</h3><p>NVIDIA has already outlined its next-generation roadmap:</p><pre>| Architecture    | Expected Timeline | Key Innovations             |<br>| --------------- | ----------------- | --------------------------- |<br>| Blackwell       | 2024              | Chiplet GPU, FP4, NVLink 5  |<br>| Blackwell Ultra | 2025              | 288 GB HBM3e                |<br>| Vera Rubin      | 2026              | HBM4, 13 TB/s bandwidth     |<br>| Rubin Ultra     | 2027              | Optical interconnects       |<br>| Feynman         | 2028              | Photonic rack-scale fabrics |</pre><p>The long-term direction is clear: GPUs are evolving into distributed AI fabrics rather than standalone accelerators.</p><h3>Conclusion</h3><p>Blackwell represents more than another GPU generation.</p><p>It is a complete rethinking of AI compute architecture:</p><ul><li>Chiplet-based GPU design overcomes physical manufacturing limits</li><li>Native FP4 makes ultra-efficient AI inference mainstream</li><li>TMEM introduces a transformer-optimized memory layer</li><li>NVLink 5 transforms racks into unified AI supercomputers</li><li>Operational features target real-world AI infrastructure requirements</li></ul><p>For AI architects, platform engineers, and MLOps teams, the shift is profound.</p><p>The industry is moving from viewing GPUs as isolated accelerators to treating them as interconnected AI infrastructure fabrics — and Blackwell is the first architecture fully designed for that future.</p><p><em>Written for AI/ML engineers, platform architects, MLOps professionals, and infrastructure teams exploring next-generation AI systems.</em></p><p>#NVIDIA #Blackwell #GPU #AIInfrastructure #LLMOps #MLOps #GenerativeAI #DeepLearning #MachineLearning #AIArchitecture</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5892679ae82a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/inside-nvidia-blackwell-the-gpu-architecture-redefining-the-age-of-ai-5892679ae82a">Inside NVIDIA Blackwell: The GPU Architecture Redefining the Age of AI</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Copy Fail: When a 9-Year-Old Linux Kernel Bug Turns Into a Real Threat]]></title>
            <link>https://medium.com/techbeatly/copy-fail-when-a-9-year-old-linux-kernel-bug-turns-into-a-real-threat-35fae763d14d?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/35fae763d14d</guid>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[security]]></category>
            <dc:creator><![CDATA[Ebin Babu]]></dc:creator>
            <pubDate>Sat, 02 May 2026 06:32:57 GMT</pubDate>
            <atom:updated>2026-05-02T06:32:55.868Z</atom:updated>
            <content:encoded><![CDATA[<p>Sometimes the most dangerous vulnerabilities aren’t the newest ones they’re the ones quietly sitting in plain sight for years. That’s exactly what happened with CVE-2026-31431, a Linux kernel flaw dating back to 2017 that has now surfaced with a public exploit.</p><blockquote>And yes it’s as serious as it sounds.</blockquote><p><strong>What Makes This Vulnerability Stand Out?</strong></p><blockquote>This isn’t your typical “hard-to-exploit” edge case. CVE-2026-31431 is a Local Privilege Escalation (LPE) vulnerability, meaning an attacker who already has any level of access can escalate privileges to root.</blockquote><p>What’s particularly concerning:</p><p>No race conditions → exploitation is stable and reliable</p><p>Works across major Linux distributions</p><p>Low-privileged user → root in seconds</p><p>Shared kernel impact → affects containers, multi-tenant systems, and shared environments</p><p>In simple terms:</p><p>&gt; If an attacker gets any foothold on your system, they can quickly gain full control of the node.</p><p>Cloud Reality Check: Who Is Responsible?</p><p>If you&#39;re running workloads on platforms like</p><ul><li>Amazon Web Services (AWS)</li><li>Microsoft Azure</li><li>Google cloud platform ( GCP)</li></ul><p>…it’s important to understand the shared responsibility model.</p><p>This vulnerability sits squarely in the customer responsibility layer.</p><p>That means:</p><p>If you’re running EC2 instances or virtual machines</p><p>If you manage your own OS and kernel</p><p>You are responsible for patching the kernel not the cloud provider.</p><p>How to Fix (And Why It Matters)</p><p>While waiting for stable kernel patches to be released, we can’t sit idle. Immediate risk reduction was critical.</p><p>1. Applied Interim Mitigation</p><p>We can block the vulnerable module at boot time:</p><p>grubby --update-kernel=ALL --args=&quot;initcall_blacklist=algif_aead_init&quot;</p><p>Important note:<br>The common modprobe.d workaround does NOT work on RHEL-based distributions, because the algif_aead component is compiled directly into the kernel not loaded as a module.</p><p>2. Preparing for Patch Rollout</p><p>Be actively checking the upstream fixes and get ready to:</p><p>Roll out kernel updates as soon as they land in stable repositories</p><p>Prioritize updates across critical workloads</p><p>3. Tightening Runtime Security</p><p>We can also reinforced our runtime posture:</p><ul><li>Enforced non-root execution wherever possible</li><li>Disabled privilege escalation paths</li><li>Reduced blast radius across workloads</li></ul><p>Why This Matters More Than Usual</p><p>This vulnerability is a perfect example of a “latent risk” turning into an active threat:</p><p>It existed silently for 9 years</p><p>It impacts core kernel behavior</p><p>It’s now weaponized and public</p><p>In modern cloud-native environments especially with containers and shared kernels the impact multiplies quickly.</p><p>Final Takeaway</p><p>Security isn’t just about reacting to new threats it’s about continuously validating old assumptions.</p><ul><li>A single low-privileged foothold should never be enough to take over your entire system.</li></ul><p>If your infrastructure relies on Linux (and most does), now is the time to:</p><p>Patch aggressively</p><p>Audit privilege boundaries</p><p>Re-evaluate kernel exposure</p><p>Because sometimes, the biggest risks are the ones that have been waiting the longest.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=35fae763d14d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/copy-fail-when-a-9-year-old-linux-kernel-bug-turns-into-a-real-threat-35fae763d14d">Copy Fail: When a 9-Year-Old Linux Kernel Bug Turns Into a Real Threat</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Getting Started with Knative Serving on AWS EKS: A Step-by-Step Guide]]></title>
            <link>https://medium.com/techbeatly/getting-started-with-knative-serving-on-aws-eks-a-step-by-step-guide-ac4067fc4c4e?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/ac4067fc4c4e</guid>
            <category><![CDATA[knative]]></category>
            <category><![CDATA[aws-eks]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[serverless]]></category>
            <category><![CDATA[knative-eventing]]></category>
            <dc:creator><![CDATA[Alexy Pulivelil]]></dc:creator>
            <pubDate>Sat, 04 Apr 2026 02:55:34 GMT</pubDate>
            <atom:updated>2026-04-04T02:55:33.093Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why I Started Looking at Knative</h3><p>In my day-to-day work on AWS EKS, I had been using <strong>HPA (Horizontal Pod Autoscaler)</strong> to scale workloads up and down based on CPU/memory metrics and <strong>kube-downscaler</strong> to shut down non-production environments outside business hours. It worked 😒 but it felt like stitching two tools together to solve a problem that should have a cleaner answer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XvNwGMtNojOlKTvSLHNUgg.png" /></figure><p>Then I came across <strong>Knative</strong>.</p><p>The promise of scale-to-zero, traffic-based autoscaling, and a clean abstraction over Kubernetes deployments caught my attention. So I decided to spin up a fresh EKS cluster and get hands-on with it. This post walks through everything I did from cluster creation to deploying my first Knative Service.</p><h3>What is Knative?</h3><p>Knative is an open-source platform built on Kubernetes that adds two major capabilities:</p><ul><li><strong>Knative Serving</strong> — Run serverless-style workloads that automatically scale up on traffic and scale down to zero when idle</li><li><strong>Knative Eventing</strong> — Build event-driven architectures on Kubernetes using CloudEvents</li></ul><p>For this post, we’re focusing entirely on <strong>Knative Serving</strong> — the component that handles deploying and scaling containerised applications.</p><p>At its core, Knative Serving introduces a new Kubernetes resource called a <strong>Knative Service (ksvc)</strong>. When you create one, Knative manages:</p><ul><li>The underlying Deployment</li><li>A <strong>Revision</strong> (an immutable snapshot of your configuration)</li><li>Traffic routing (you can split traffic across revisions for canary releases)</li><li>Autoscaling — including all the way down to zero replicas</li></ul><h3>Scale-to-Zero: The Core Magic🧙‍♂️</h3><p>With standard Kubernetes + HPA, your pods scale based on CPU or memory thresholds — but they never go to zero. You always have at least one pod running, even when there’s no traffic. That costs money💰.<br>Knative Serving scales based on <strong>actual incoming requests</strong>. When there are no requests, it scales your pods to zero. When a request comes in, the <strong>Activator</strong> component buffers it, wakes up a pod, and forwards the request all in seconds.<br>For non-production environments, dev/qa clusters, or bursty workloads, this is a significant cost saver.</p><h3>Prerequisites</h3><ul><li>AWS CLI configured with AWS Access Key and Secret Key</li><li>eksctl installed (<a href="https://eksctl.io/installation/">installation guide</a>)</li><li>kubectl installed</li><li>Basic familiarity with Kubernetes</li></ul><p>Step 1: Spin Up an EKS Cluster with eksctl</p><p>One of the best things about eksctl is that it creates the entire AWS networking stack for you — VPC, public and private subnets across two AZs, Internet Gateway, NAT Gateway, route tables, and the correct subnet tags for EKS. No manual VPC setup needed.</p><pre>eksctl create cluster \<br>  --name knative-dev \<br>  --region ap-south-1 \<br>  --nodegroup-name standard-workers \<br>  --node-type t3.medium \<br>  --nodes 2 \<br>  --nodes-min 1 \<br>  --nodes-max 3 \<br>  --managed \<br>  --profile &lt;your-aws-profile&gt;</pre><p>This takes around 15 minutes. Once done, your kubeconfig is automatically updated and you&#39;re ready to run kubectl commands.<br><strong>Cost tip:</strong> Use a single NAT Gateway (eksctl default) for dev setups. Using one per AZ costs ~$32/month extra — unnecessary for learning.</p><p>Step 2: Install Knative Serving<br>Apply the CRDs</p><pre>kubectl apply -f https://github.com/knative/serving/releases/download/knative-v1.13.0/serving-crds.yaml</pre><p>Install the Core Components</p><pre>kubectl apply -f https://github.com/knative/serving/releases/download/knative-v1.13.0/serving-core.yaml</pre><p>Install Kourier (Ingress)<br>Knative needs an ingress layer to route external traffic to your services. <strong>Kourier</strong> is the lightweight, officially supported option — ideal for getting started without the overhead of Istio.</p><pre>kubectl apply -f https://github.com/knative/net-kourier/releases/download/knative-v1.13.0/kourier.yaml</pre><p>Tell Knative to use Kourier as its ingress:</p><pre>kubectl patch configmap/config-network \<br>  --namespace knative-serving \<br>  --type merge \<br>  --patch &#39;{&quot;data&quot;:{&quot;ingress-class&quot;:&quot;kourier.ingress.networking.knative.dev&quot;}}&#39;</pre><p>Step 4: Configure DNS with sslip.io<br>When Kourier spins up on EKS, it creates an AWS Load Balancer with a hostname like:</p><pre>a1b2c3d4.ap-south-1.elb.amazonaws.com</pre><p>For Knative Services to get usable URLs, we need DNS. For a dev setup, <strong>sslip.io</strong> is a magic DNS service that resolves any hostname containing an IP address back to that IP. So your service gets a URL like:</p><pre>http://hello.default.52.66.12.34.sslip.io</pre><p>Step 5: Verify Everything is Running</p><pre>kubectl get pods -n knative-serving<br>kubectl get pods -n kourier-system</pre><p>You should see all pods in Running state</p><p>Here’s what each component does:</p><p><strong>activator </strong>Buffers requests when pods are at zero, wakes them up</p><p><strong>autoscaler </strong>Watches traffic metrics and decides when to scale</p><p><strong>controller </strong>Reconciles Knative resources with Kubernetes</p><p><strong>webhook </strong>Validates Knative YAML before it’s applied</p><p><strong>net-kourier-controller </strong>Manages Envoy routing rules for Kourier</p><p><strong>scale-kourier-gateway </strong>The actual ingress gateway receiving traffic</p><p>Step 6: Deploy Your First Knative Service</p><pre>apiVersion: serving.knative.dev/v1<br>kind: Service<br>metadata:<br>  name: hello<br>  namespace: default<br>spec:<br>  template:<br>    spec:<br>      containers:<br>        - image: gcr.io/knative-samples/helloworld-go<br>          env:<br>            - name: TARGET<br>              value: &quot;Knative on EKS&quot;</pre><p>Apply it: <em>kubectl apply -f hello-ksvc.yaml</em></p><p>Step 7: Test Scale-to-Zero<br>Now wait 60–90 seconds and watch the pods<br>You’ll see the pod terminate — scaled to zero. No traffic, no running pods.</p><p>Hit the URL again — the Activator catches the request, spins up a new pod, and you get your response back. That’s the entire scale-to-zero loop in action.</p><h3>Cost Saving Angle</h3><p>Here’s how this compares to my previous setup:</p><ul><li>HPA + kube-downscaler — 1 pod (during hours), 0 outside hours</li><li>Knative Serving — 0 pods</li></ul><p>With Knative, you don’t need a separate tool to shut things down. It handles both the scale-up and the scale-to-zero automatically. For dev/staging workloads that sit idle between deployments or tests, this translates to real savings on EC2 compute.</p><p>NB: EKS control plane alone costs ~$0.10/hr — always clean up dev clusters.</p><pre>eksctl delete cluster \<br>  --name knative-dev \<br>  --region ap-south-1 \<br>  --profile &lt;your-aws-profile&gt;</pre><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ac4067fc4c4e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/getting-started-with-knative-serving-on-aws-eks-a-step-by-step-guide-ac4067fc4c4e">Getting Started with Knative Serving on AWS EKS: A Step-by-Step Guide</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[vLLM]]></title>
            <link>https://medium.com/techbeatly/vllm-ab860850a70a?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/ab860850a70a</guid>
            <dc:creator><![CDATA[Praveenpm]]></dc:creator>
            <pubDate>Sun, 22 Mar 2026 16:25:48 GMT</pubDate>
            <atom:updated>2026-03-22T16:29:10.650Z</atom:updated>
            <content:encoded><![CDATA[<p>I’ve been digging into vLLM’s. Here’s what actually makes it fast — from the GPU level up.</p><p>The problem no one talks about: Before vLLM, LLM servers wasted ~75.6% of GPU memory. Not a typo. Internal fragmentation (~20%) + external fragmentation (~55%) from pre-allocated KV caches meant you were paying for 5 GPUs to do the work of 1.</p><p>PagedAttention fixed this by applying OS virtual memory paging to KV cache. Physical GPU blocks are mapped via a per-sequence block table. Lazy allocation. Immediate free on completion. Copy-on-write for shared prefixes. The result: near-zero memory waste.</p><p>Continuous batching eliminated the other bottleneck — GPU idle time between batches. Instead of waiting for the whole batch to finish, vLLM schedules at the iteration level. New requests join the moment a slot opens. GPU utilisation goes from ~40% to ~95%.</p><pre>The numbers on an A100:<br>— vLLM: 17.12 req/s (OPT-13B, ShareGPT)<br>— FasterTransformer: 1.07 req/s<br>— HuggingFace TGI: 0.71 req/s<br>— Orca static: 0.59 req/s</pre><p>And it scales: 8× A100s with tensor parallelism → ~7.2× throughput. Cross-node pipeline parallelism supported via InfiniBand.</p><p>Quantisation (INT4/AWQ) lets you run LLaMA-70B on a single A100 80GB — which normally requires two cards in FP16.</p><p>The client layer is a drop-in OpenAI replacement. Change base_url, nothing else.</p><pre>See the below PPT</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/853/1*4DhXDFvhmpef3YgKJP8yhQ.gif" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab860850a70a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/vllm-ab860850a70a">vLLM</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Best Ways to Set Up an Ansible Development Environment (Linux, Windows, macOS, and Windows WSL)]]></title>
            <link>https://medium.com/techbeatly/best-ways-to-set-up-an-ansible-development-environment-linux-windows-macos-and-windows-wsl-1791b472b2eb?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/1791b472b2eb</guid>
            <category><![CDATA[ansible]]></category>
            <category><![CDATA[developer]]></category>
            <dc:creator><![CDATA[Gineesh Madapparambath]]></dc:creator>
            <pubDate>Fri, 06 Mar 2026 04:20:30 GMT</pubDate>
            <atom:updated>2026-03-06T04:21:02.226Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hXRmoTWB9S-5SbUS.png" /></figure><p>If you want to write Ansible playbooks, roles, modules, or collections, the first step is a good development environment.</p><p><em>This story was originally published at </em><a href="https://techbeatly.com/ansible-development-environment/"><em>techbeatly.com</em></a><em>.</em></p><p>The good news is that you have many options. Some are simple and run directly on your laptop. Others are better for teams or enterprise environments.</p><p>In this guide, we walk through practical ways to build an Ansible development and testing setup. These options work for Linux, macOS, and Windows users.</p><p>Many Windows users struggle to run Ansible locally. So we also cover the easiest ways for Windows users to build a working environment.</p><p>Let’s go through the options.</p><h3>1. Local Linux or macOS Environment (Simple and Fast)</h3><p>If you already use Linux or macOS, this is usually the easiest option.</p><p>You <a href="https://docs.ansible.com/projects/ansible/latest/installation_guide/intro_installation.html">install Ansible</a> and its tools directly on your system and start developing immediately.</p><p>This gives the best performance because everything runs locally.</p><h3>Requirements</h3><ul><li>Linux or macOS machine</li><li>Python installed</li><li>Ability to install packages (pip or system package manager)</li></ul><h3>Typical Setup</h3><p>Install Ansible using pip:</p><pre>python3 -m pip install --user ansible</pre><p>Or use the package manager:</p><p>Ubuntu / Debian</p><pre>sudo apt install ansible</pre><p>Fedora</p><pre>sudo dnf install ansible</pre><p>Once installed, you can start writing playbooks, roles, and collections.</p><p>This setup works well if you already use Linux for daily work.</p><h3>2. Virtual Machine Lab (Recommended for Windows Users)</h3><p>If you use Windows or want a clean Linux environment, a virtual machine is a great choice.</p><p>You create a Linux VM and run Ansible inside it. Your laptop acts as the host, and the VM becomes your Ansible lab.</p><p>Popular VM tools:</p><ul><li>VirtualBox</li><li>VMware Workstation</li><li>Hyper-V</li></ul><p>Typical Linux images used for Ansible labs:</p><ul><li>Ubuntu</li><li>Fedora</li><li>CentOS Stream</li><li>Rocky Linux</li><li>AlmaLinux</li></ul><p>You can connect to the VM using:</p><ul><li>SSH</li><li>VS Code Remote SSH</li><li>Terminal</li></ul><p>This approach keeps your host system clean and provides a consistent Linux environment.</p><h3>Example Lab Guide</h3><ul><li><a href="https://techbeatly.com/now-available-updated-ansible-lab-using-vagrant-and-virtualbox/">Now Available: Updated Ansible Lab using Vagrant and VirtualBox!</a></li><li><a href="https://www.youtube.com/watch?v=qoliqxGvX84">Setting up Lab Environment for Ansible | VirtualBox, Vagrant</a></li></ul><h3>3. Vagrant-Based Ansible Lab</h3><p>If you regularly build and destroy labs, Vagrant is extremely useful.</p><p>Vagrant allows you to define your entire lab environment as code.</p><p>For example, you can automatically create:</p><ul><li>An Ansible control node</li><li>Multiple managed nodes</li><li>Preconfigured networking</li><li>Preinstalled tools</li></ul><p>With one command, the whole lab can start.</p><pre>vagrant up</pre><p>And you can destroy it just as easily.</p><pre>vagrant destroy</pre><p>This approach is great for:</p><ul><li>learning Ansible</li><li>testing playbooks</li><li>building repeatable labs</li><li>training environments</li></ul><p>Example setup:</p><ul><li><a href="https://techbeatly.com/now-available-updated-ansible-lab-using-vagrant-and-virtualbox/">Now Available: Updated Ansible Lab using Vagrant and VirtualBox!</a></li></ul><h3>4. Windows Subsystem for Linux (WSL)</h3><p>Ansible does not run natively on Windows, but Windows Subsystem for Linux (WSL) solves this problem by allowing you to run a full Linux environment directly inside Windows. From the user’s perspective, it behaves almost like a native Linux machine, which makes it a very practical way for Windows users to install and run Ansible without needing a full virtual machine.</p><h3>Requirements</h3><ul><li>Windows 10 or Windows 11</li><li>WSL enabled</li><li>A Linux distribution from the Microsoft Store</li></ul><p>Common choices:</p><ul><li>Ubuntu</li><li>Debian</li></ul><p>Once installed, you can install Ansible normally inside WSL.</p><p>Example:</p><pre>sudo apt update<br>sudo apt install ansible</pre><p>Many developers combine WSL with VS Code for a very smooth workflow.</p><h3>5. VS Code Remote Development</h3><p>Another popular approach is to develop locally while running Ansible on a remote Linux server using the VS Code Remote SSH extension. In this setup, your code editor runs on your local machine, but the files and Ansible execution happen on the remote system, giving you the benefits of a powerful Linux environment without installing everything on your laptop.</p><p>Advantages:</p><ul><li>No need to install tools locally</li><li>Works on Windows, Linux, and macOS</li><li>Remote server provides consistent environment</li><li>Easy collaboration for teams</li></ul><p>This method is common in corporate environments where developers connect to shared Linux servers.</p><p>Detailed guide:</p><ul><li><a href="https://techbeatly.com/ansible-and-vscode-remote-development/">Ansible and VSCode Remote Development</a></li></ul><h3>6. Cloud-Based Ansible Lab</h3><p>If your laptop is limited or you want a disposable environment, a cloud lab is another good option.</p><p>You can create temporary infrastructure in a public cloud and run your Ansible control node there.</p><p>Typical setup:</p><ul><li>Ansible control node (EC2 instance or VM)</li><li>Several test servers</li><li>SSH connectivity</li></ul><p>This lab can be destroyed after testing.</p><p>You can automate the entire lab creation using Terraform.</p><p>Example guides:</p><p>Create a free lab in public cloud:</p><ul><li><a href="https://techbeatly.com/creating-a-free-ansible-lab-in-public-cloud/">How to Create a FREE Ansible Lab in Public Cloud (AWS, GCP, Azure)</a></li><li><a href="https://www.techbeatly.com/2021/06/use-terraform-to-create-a-free-ansible-lab-in-aws.html">Use Terraform to Create a FREE Ansible Lab in AWS</a></li></ul><p>This approach is great for:</p><ul><li>testing infrastructure automation</li><li>learning cloud automation</li><li>CI/CD experiments</li></ul><h3>7. Container-Based Development (Advanced Option)</h3><p>Another modern option is running your Ansible environment inside containers.</p><p>Tools often used:</p><ul><li>Podman</li><li>Docker</li><li>Dev Containers</li></ul><p>This allows you to package the entire development environment as a container image.</p><p>Advantages:</p><ul><li>reproducible environment</li><li>easy onboarding for teams</li><li>same environment for everyone</li></ul><p>Many teams use this method together with VS Code Dev Containers.</p><h3>8. OpenShift Dev Spaces (Enterprise Teams)</h3><p>Large teams often need a centralized and standardized development platform.</p><p><a href="https://developers.redhat.com/products/openshift-dev-spaces">OpenShift Dev Spaces</a> provides a web-based development environment running inside an OpenShift cluster.</p><p>The full development stack runs inside the cluster.</p><p>Developers only need:</p><ul><li>a browser</li><li>access to the Dev Spaces environment</li></ul><p>Benefits:</p><ul><li>identical environment for every developer</li><li>no local setup required</li><li>easier management for organizations</li><li>integrated with enterprise security and identity systems</li></ul><p>This option is common in large enterprise environments.</p><h3>Final Thoughts</h3><p>There is no single “best” Ansible development environment. The right choice depends on several factors including your operating system, available hardware resources, team requirements, and any enterprise policies you need to follow. What really matters is having a clean and repeatable environment where you can safely test playbooks and automation. Once that foundation is in place, developing Ansible content becomes much easier. If you are just starting out, the simplest approach is to use WSL or a small Linux virtual machine with Ansible installed, and later expand your setup into cloud labs, container-based environments, or enterprise platforms as your needs grow.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1791b472b2eb" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/best-ways-to-set-up-an-ansible-development-environment-linux-windows-macos-and-windows-wsl-1791b472b2eb">Best Ways to Set Up an Ansible Development Environment (Linux, Windows, macOS, and Windows WSL)</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Testing Kubernetes v1.35.0 with minikube: A Step-by-Step Guide for multi-node cluster]]></title>
            <link>https://medium.com/techbeatly/testing-kubernetes-v1-35-0-with-minikube-a-step-by-step-guide-for-multi-node-cluster-d5d8317da0ff?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/d5d8317da0ff</guid>
            <category><![CDATA[minikube]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Gineesh Madapparambath]]></dc:creator>
            <pubDate>Sat, 06 Dec 2025 15:26:26 GMT</pubDate>
            <atom:updated>2025-12-06T15:26:25.341Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1EK6XmKbLWMaob7_RIYzUA.png" /></figure><p>In every Kubernetes release cycle, the release candidates (RCs) play a big role. They help maintainers, contributors, platform engineers, and curious folks like us validate upcoming changes before the final release drops. Testing RC builds early helps catch regressions, check integrations, and update documentation or automation before production adopters hit the new version.</p><p><em>This article was originally published in </em><a href="https://techbeatly.com/kubernetes-1-35-with-minikube/"><strong><em>techbeatly</em></strong></a><em>.</em></p><p>Kubernetes 1.35 is almost here, with general availability landing on 17 December 2025. The release cycle kicked off in mid-September and has already passed all major freeze milestones, and with the RC out now, it’s a good time to spin up a local cluster and try the new features before GA.</p><h3>Prerequisites</h3><ul><li>Minikube installed</li><li>VirtualBox (or Podman/Docker)</li><li>Basic Kubernetes knowledge</li><li>Your user-installed minikube binary is located at: ~/.local/bin/minikube</li></ul><h3>Update Minikube First</h3><p>Check the current version:</p><pre>$ minikube update-check</pre><p>If you see something like:</p><pre>$ minikube update-check<br>CurrentVersion: v1.35.0<br>LatestVersion: v1.37.0</pre><p>Update your user-local binary:</p><pre>$ curl -Lo ~/.local/bin/minikube   https://storage.googleapis.com/minikube/releases/latest/minikube-linux-amd64<br>$ chmod +x ~/.local/bin/minikube</pre><p>Verify:</p><pre>$ minikube version</pre><h3>Start a Multi-Node Kubernetes v1.35.0-rc.0 Cluster (VirtualBox)</h3><pre>$ minikube start \<br>    --driver=podman \<br>    --nodes 3 \<br>    --cni calico \<br>    --cpus=2 \<br>    --memory=2g \<br>    --kubernetes-version=v1.35.0-rc.0 \<br>    --container-runtime=containerd \<br>    --profile k8s-1-35-rc</pre><h3>What these flags mean</h3><ul><li>--nodes 3: three-node cluster, 1st node will become the controller/master node.</li><li>--driver=podman: use <a href="https://techbeatly.com/podman-rhel9/">Podman</a> container as Kubernetes nodes. Use --driver=virtualbox if you want to use VirtualBox VMs as the backend</li><li>--cni calico → networking via Calico</li><li>--profile k8s-1-35-rc → keeps this cluster separate from others</li><li>--kubernetes-version=v1.35.0-rc.0 → explicitly test the RC release</li></ul><p>As Minikube starts, you’ll see logs confirming RC lookup, node creation, control plane boot, CNI setup, and worker node join workflow.</p><h3>Verify the Cluster</h3><pre>$ kubectl get nodes<br>NAME              STATUS   ROLES           AGE     VERSION<br>k8s-1-35-rc       Ready    control-plane   2m34s   v1.35.0-rc.0<br>k8s-1-35-rc-m02   Ready    &lt;none&gt;          2m12s   v1.35.0-rc.0<br>k8s-1-35-rc-m03   Ready    &lt;none&gt;          117s    v1.35.0-rc.0</pre><p>At this point you have a full multi-node Kubernetes v1.35 RC cluster running locally.</p><h3>Managing Multiple Minikube Profiles</h3><p>List all profiles:</p><pre>$ minikube profile list</pre><p>Switch profiles:</p><pre>$ minikube profile k8s-1-35-rc<br>✅  minikube profile was successfully set to k8s-1-35-rc</pre><p>Stop a cluster:</p><pre>$ minikube stop --profile=k8s-1-35-rc</pre><p>Start again if needed:</p><pre>$ minikube start --profile=k8s-1-35-rc</pre><p>Delete once testing finished</p><pre>$ minikube delete --profile=k8s-1-35-rc</pre><h3>Wrap-up</h3><p>You now have a repeatable way to test Kubernetes v1.35.0-rc.0 and other versions using Minikube, either with Podman or VirtualBox.This is the ideal setup for trying new features, checking compatibility, and preparing for the final GA release in December 2025.</p><p><em>This article was originally published in </em><a href="https://techbeatly.com/kubernetes-1-35-with-minikube/"><strong><em>techbeatly</em></strong></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5d8317da0ff" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/testing-kubernetes-v1-35-0-with-minikube-a-step-by-step-guide-for-multi-node-cluster-d5d8317da0ff">Testing Kubernetes v1.35.0 with minikube: A Step-by-Step Guide for multi-node cluster</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Save 30% on KubeCon + CloudNativeCon North America 2025 Tickets!]]></title>
            <link>https://medium.com/techbeatly/save-30-on-kubecon-cloudnativecon-north-america-2025-tickets-394f7475ca5b?source=rss----ff49a217e60---4</link>
            <guid isPermaLink="false">https://medium.com/p/394f7475ca5b</guid>
            <category><![CDATA[the-linux-foundation]]></category>
            <category><![CDATA[cloudnativecon]]></category>
            <category><![CDATA[kubeconna]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[kubecon]]></category>
            <dc:creator><![CDATA[Gineesh Madapparambath]]></dc:creator>
            <pubDate>Tue, 22 Jul 2025 02:13:31 GMT</pubDate>
            <atom:updated>2025-07-22T03:14:59.601Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*pkzZF5Ofl6KoVzgI.png" /></figure><p>Are you into Kubernetes, cloud native tech, or open source in general? Then mark your calendar:<a href="https://tbly.cc/kubecon-na"> <strong>KubeCon + CloudNativeCon North America 2025</strong></a> is happening from <strong>November 10–13, 2025</strong> in <strong>Atlanta, Georgia</strong> — and it’s going to be big.</p><p><em>This article was originally published in </em><a href="https://www.techbeatly.com/kubecon-cloudnativecon-north-america-2025/"><strong><em>techbeatly</em></strong></a><em>.</em></p><p>This is <strong>CNCF’s flagship event</strong>, bringing together adopters, contributors, and cloud native technologists from around the world. Whether you’re building platforms, running production workloads, contributing to open source, or just curious about what’s next — this is where it’s all happening.</p><h3>Here’s what to expect:</h3><ul><li><strong>Deep Dives &amp; Learning Tracks</strong>: Kubernetes, containers, security, observability, platform engineering, AI, and more — all covered in technical sessions, workshops, and hands-on demos.</li><li><strong>Networking &amp; Hiring</strong>: Meet peers, project maintainers, hiring managers, and leaders from the top cloud native companies.</li><li><strong>Keynotes from Industry Leaders</strong>: Hear directly from the folks building and shaping the cloud native ecosystem.</li><li><strong>Project Updates</strong>: Get the latest from CNCF’s Graduated, Incubating, and Sandbox Projects.</li><li><strong>Co-Located Events</strong>: Dive into specialized topics with focused events running alongside the main conference.</li></ul><h3>Special Offer: 30% Off Current Pricing!</h3><p><a href="https://tbly.cc/kubecon-na"><strong>techbeatly</strong></a> has an <strong>exclusive 30% discount</strong> for you. Just use <strong>promo code: TECHBEATLY</strong> when registering.</p><p>I<strong>mportant</strong>: This discount <strong>expires on August 26, 2025</strong> — don’t wait too long.</p><p><a href="https://tbly.cc/kubecon-na"><strong>Register now</strong></a> and lock in your ticket at a lower price!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/884/0*6NokGYjQej7SRnji.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/886/0*bHtn41zMWyAnzuTs.png" /></figure><h3>Why You Shouldn’t Miss This:</h3><ul><li>Stay ahead in the cloud native space</li><li>Learn directly from contributors and industry experts</li><li>Build your network and meet potential collaborators or employers</li><li>Experience the energy of the global open source community</li></ul><p>Whether you’re just getting started or knee-deep in production, this is the one <a href="https://tbly.cc/kubecon-na"><strong>event</strong></a> that’ll keep you informed, inspired, and connected.</p><p><em>This article was originally published in </em><a href="https://www.techbeatly.com/kubecon-cloudnativecon-north-america-2025/"><strong><em>techbeatly</em></strong></a><em>.</em></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FPHsNO6A2EaU%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DPHsNO6A2EaU&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FPHsNO6A2EaU%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/2fb2121171926d0b32d37ebeffff1125/href">https://medium.com/media/2fb2121171926d0b32d37ebeffff1125/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=394f7475ca5b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/techbeatly/save-30-on-kubecon-cloudnativecon-north-america-2025-tickets-394f7475ca5b">Save 30% on KubeCon + CloudNativeCon North America 2025 Tickets!</a> was originally published in <a href="https://medium.com/techbeatly">techbeatly</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>