<?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 Michal Golda on Medium]]></title>
        <description><![CDATA[Stories by Michal Golda on Medium]]></description>
        <link>https://medium.com/@mjgolda?source=rss-2aee0228c8a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*adUcrOLSQNw5bSsEG3AWyQ.png</url>
            <title>Stories by Michal Golda on Medium</title>
            <link>https://medium.com/@mjgolda?source=rss-2aee0228c8a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 16:36:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mjgolda/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[Demystifying the Black Box: How the New GKE Autoscaling Experience Builds Trust and Efficiency]]></title>
            <link>https://medium.com/@mjgolda/demystifying-the-black-box-how-the-new-gke-autoscaling-experience-builds-trust-and-efficiency-4ac3e4ce290f?source=rss-2aee0228c8a------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ac3e4ce290f</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[kubernetes-autoscaling]]></category>
            <category><![CDATA[gke]]></category>
            <dc:creator><![CDATA[Michal Golda]]></dc:creator>
            <pubDate>Thu, 12 Mar 2026 10:02:01 GMT</pubDate>
            <atom:updated>2026-03-12T10:02:01.252Z</atom:updated>
            <content:encoded><![CDATA[<p>Text written by <a href="https://medium.com/u/99508bd53ae5">Ewelina Skłodowska</a> and <a href="https://medium.com/u/2aee0228c8a">Michal Golda</a></p><p>Autoscaling offers the best of both worlds: high availability and cost efficiency. Yet, despite high adoption rates, many teams hesitate to fully automate their critical workloads. Why?</p><p><strong>Because autoscaling often feels like a “black box”</strong></p><p>When scaling behavior or costs shift unexpectedly, you need deep visibility to maintain operational confidence. To solve this, we are introducing a unified Scaling view within GKE. By consolidating configuration and observability into a central workspace, we enable you to manage and optimize operations without the constant context switching.<br>Here is how the new experience tackles the three biggest friction points in scaling.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-pEt1ZQaXTR_FicopelutQ.png" /></figure><h3>Answering “Why did (or didn’t) it scale?”</h3><p><strong>The Challenge:</strong> One of the most frequent support questions we see is: “Why did my workload fail to scale?”. Answering this usually requires digging through disparate logs to correlate events with resource usage.</p><p><strong>The Solution:</strong> We’ve moved the data out of the logs and onto your charts. The new view overlays scaling decisions directly onto your metrics, giving you immediate context.</p><p><em>Drill Down</em>: If a high-traffic pod doesn’t scale, you can investigate immediately. Move from workload-level averages down to container-level utilization to spot the specific outlier.</p><p><em>Inspect the Logic</em>: See the “why” behind the “what.” View HPA targets alongside VPA recommendations, and inspect internal logic — such as dampening or calculation thresholds — to understand exactly why the system decided to hold steady.</p><p><em>Spot Conflicts</em>: Running HPA and VPA together often feels like a tug-of-war. The new view visualizes their interaction, allowing you to instantly spot “thrashing” — where VPA resizing conflicts with HPA scaling — so you can tune them to work in harmony.</p><p><em>Clarify with Context</em>: With clearer, human-readable HPA logs, you can diagnose behavior without deciphering raw text, giving you the immediate context you need to troubleshoot.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZTbolOZqoeKs_CIXjejdqw.png" /></figure><h3>One-Click Configuration with Best Practices</h3><p><strong>The Challenge:</strong> Translating high-level application behavior into precise YAML is often a time-consuming guessing game.</p><p><strong>The Solution:</strong> A unified configuration flow designed for clarity. We have streamlined the setup process by introducing Scaling Presets as actionable starting points.</p><p>Avoid starting from scratch by selecting a baseline profile that matches your specific architecture. Leverage Horizontal or Vertical Pod autoscaling presets, which come with optimized default configurations that you can easily customize as needed.</p><p><em>Developer-Friendly</em>: Engineers can verify and copy the exact YAML configuration using the “Equivalent code” view, rather than relying solely on UI abstractions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J2kkVTmZmMB1C8h7N6Q8iQ.png" /></figure><p>This design lays the foundation for deeper control — paving the way for advanced metric configurations and customizable thresholds coming in future releases.</p><h3>Proactive Cost and Performance Rightsizing</h3><p><strong>The Challenge:</strong> Fear of downtime often drives up cloud bills. Users frequently configure resources significantly higher than peak usage to create a “safety buffer,” leading to wasted expense. Pinpointing this waste without compromising reliability typically requires analyzing weeks of historical data.</p><p><strong>The Solution:</strong> We are replacing manual analysis with a “Guided Rightsizing” wizard that brings active recommendations directly into your workflow.</p><p><em>Detect Risks, Not Just Costs</em>: The wizard goes beyond simple cost cutting. It automatically detects critical under-provisioning risks, identifying workloads in danger of CPU throttling or OOM (Out-of-Memory) crashes, alongside opportunities to trim wasted resources.</p><p><em>No More Financial Surprises</em>: Before you commit to a change, the integrated Cost Impact Analysis estimates the result, showing you exactly how the new configuration is projected to affect your monthly cloud bill.</p><h3>Conclusion: Turning the “Black Box” into a Glass Box</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3GhR1Qa_XQ6VvkTGsL1wug.png" /></figure><p>By consolidating configuration, observability, and optimization into a single workspace, we are transforming autoscaling from a mystery into a transparent engine.</p><p>Whether you are troubleshooting a stalled deployment or rightsizing a critical workload for cost efficiency, the new GKE Scaling experience ensures you have the visibility and control required to scale with absolute confidence.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ac3e4ce290f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>