<?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 Kaushik Mukherjee on Medium]]></title>
        <description><![CDATA[Stories by Kaushik Mukherjee on Medium]]></description>
        <link>https://medium.com/@kmukherjee?source=rss-d1b59b115e6d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*WDGrXa-EY58YVTFk</url>
            <title>Stories by Kaushik Mukherjee on Medium</title>
            <link>https://medium.com/@kmukherjee?source=rss-d1b59b115e6d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 12:16:18 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kmukherjee/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[Running a “Quiet” Liberty Server]]></title>
            <link>https://medium.com/@kmukherjee/running-a-quiet-liberty-server-7e33ec3a1363?source=rss-d1b59b115e6d------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e33ec3a1363</guid>
            <category><![CDATA[java]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[liberty]]></category>
            <category><![CDATA[websphere]]></category>
            <category><![CDATA[containers]]></category>
            <dc:creator><![CDATA[Kaushik Mukherjee]]></dc:creator>
            <pubDate>Tue, 04 Aug 2026 19:22:42 GMT</pubDate>
            <atom:updated>2026-08-05T17:51:09.372Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/264/1*6IBl5XbfwWCpbIQYOf7e6A.png" /></figure><p>An application server should be seen and not heard, especially in a containerized or serverless environment. When running in a cloud-native platform, it’s important to be mindful of background processes that consume CPU cycles, use memory, or generate unnecessary file system activity. Even small amounts of overhead can add up at scale, particularly when applications are deployed across hundreds of pods.</p><p>In this article, I’ll share a few configuration changes that can help make your Liberty server as quiet and efficient as possible. By disabling monitoring features that are primarily intended for development or dynamic administration, you can reduce unnecessary polling and align your runtime with immutable infrastructure and GitOps principles.</p><ol><li><strong>Disable the Dropins Directory</strong></li></ol><p>The dropins directory is a convenience feature that allows Liberty to automatically discover and deploy applications by monitoring a specific folder for new .war files.</p><p>While this is extremely useful during development, it provides little value in production environments where application deployment should be managed through a CI/CD pipeline. Continuous monitoring of the dropins directory creates unnecessary file system activity and can introduce a security risk if an unauthorized user gains access to the container&#39;s file system.</p><pre>&lt;applicationMonitor dropinsEnabled=&quot;false&quot; /&gt;</pre><p>2. <strong>Disable Keystore Polling</strong></p><p>By default, Liberty periodically checks whether SSL certificates, keystores, or truststores have changed on disk. This allows certificate updates to be detected without restarting the server.</p><p>For example, when certificates are mounted from a Kubernetes Secret, Liberty can automatically discover updated certificate files. However, this convenience comes with continuous polling overhead. In environments where pod restarts are the preferred mechanism for picking up certificate changes, the monitoring can be safely disabled.</p><pre>&lt;keyStore id=&quot;defaultKeyStore&quot; updateTrigger=&quot;mbean&quot; /&gt;</pre><p>3. <strong>Disable Dynamic Configuration Updates</strong></p><p>Liberty can monitor configuration files such as server.xml and trace configuration files, allowing changes to be applied without restarting the server.</p><p>While convenient for troubleshooting and development, dynamic configuration updates are generally not desirable in production containers. Allowing running containers to change behavior independently can lead to configuration drift and undermine GitOps practices that treat deployed artifacts as immutable.</p><pre>&lt;configMonitor updateTrigger=&quot;disabled&quot; /&gt;</pre><p><strong>Example Configuration</strong></p><p>The following server.xml snippet configuration disables the major polling and monitoring behaviors.</p><pre>&lt;!-- Disable application polling and dropins monitoring --&gt;<br>&lt;applicationMonitor <br>    updateTrigger=&quot;mbean&quot; <br>    pollingRate=&quot;0&quot;<br>    dropinsEnabled=&quot;false&quot; /&gt;<br><br>&lt;!-- Disable dynamic configuration updates --&gt;<br>&lt;configMonitor updateTrigger=&quot;disabled&quot; /&gt;<br><br>&lt;!-- Disable keystore polling --&gt;<br>&lt;keyStore<br>    id=&quot;defaultKeyStore&quot;<br>    password=&quot;${keystorePassword}&quot;<br>    updateTrigger=&quot;mbean&quot; /&gt;</pre><p><strong>Why This Matters for Scaling to Zero</strong></p><p>When using KNative or other serverless platforms, applications may frequently scale down to zero instances and then rapidly start new pods when traffic arrives. In these environments, every millisecond spent on initialization matters. The benefits may seem small on a single instance, but they become meaningful when multiplied across large Kubernetes clusters. Reducing unnecessary activity can lead to:</p><ul><li>Lower CPU consumption</li><li>Reduced cloud infrastructure costs</li><li>Better alignment with immutable infrastructure principles</li></ul><p>The recommendations in this article focus on reducing some of Liberty’s most common background monitoring and polling activities. While these changes can help create a quieter and more efficient runtime, they represent only a subset of the best practices for running Liberty in containers and cloud-native environments.</p><p>There are many additional areas to consider, including startup optimization, JVM tuning, container resource management, logging strategies, health checks, and Kubernetes-specific configurations. In future posts, I’ll explore more techniques for optimizing Liberty for cloud-native platforms, helping you reduce resource consumption, improve startup performance, and get the most value from your container deployments.</p><p>If a feature exists primarily to detect runtime changes, and your operational model relies on redeployment rather than mutation, consider disabling it. The less time Liberty spends watching files, scanning directories, and polling for changes, the more time it can spend doing what it was deployed to do: serving applications.</p><p>A quieter Liberty server is a more efficient Liberty server.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e33ec3a1363" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to get started with IBM AIOPs]]></title>
            <link>https://medium.com/@kmukherjee/how-to-get-started-with-ibm-aiops-78a951b92ad2?source=rss-d1b59b115e6d------2</link>
            <guid isPermaLink="false">https://medium.com/p/78a951b92ad2</guid>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[aiops]]></category>
            <category><![CDATA[foundation-models]]></category>
            <category><![CDATA[watson]]></category>
            <dc:creator><![CDATA[Kaushik Mukherjee]]></dc:creator>
            <pubDate>Sat, 17 Jun 2023 17:02:29 GMT</pubDate>
            <atom:updated>2023-06-20T18:33:47.954Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n3xX6UkYm60y4HqUpsSTOA.jpeg" /></figure><p>Over the last few years the IBM team has worked with hundreds of different customers on the benefits of implementing AIOPs as part of their IT operations. We have seen some amazing transformations after deploying IBM’s AIOPs solutions, but the initial adoption of any new technology can seem daunting and often bring more questions than answers. I’ve taken many of the common questions and concerns I’ve heard across multiple customers and answered them below.</p><ol><li><strong>Do we need to replace all the current tools/processes in order to adopt IBM’s AIOPs?</strong></li></ol><ul><li>Not at all. If you select a AIOps platform that is domain (and vendor) agnostic and is designed to connect to and understand the increasingly diverse datasets across multiple use cases.</li><li>For example, IBM’s approach to AIOPs is to utilize your existing tools as data sources and tie that data together in a meaningful way across the entire IT stack from the application to all the elements of the underlying infrastructure. If there are gaps or deficiencies with your current tools, then it makes sense to supplement or replace specific tools in that case.</li></ul><p><strong>2. What changes do we need to make to our internal processes and workflows to implement an AIOPs solution?</strong></p><ul><li>This is an important question to address since AIOPs is more than a set of tools or technologies. One of the important changes to make in adopting an AIOPs solution is to align teams to directly impact customer satisfaction. One example of this is to build Site Reliability Engineering teams (SRE) that are vertically integrated across IT domains to support particular lines of business. This type of structure incents domain experts to work together to resolve issues that are affecting end users.</li><li>Operations teams often have a multitude of tools providing insights, but are overwhelmed by what actions to take and the impact their actions may have. An important aspect of AIOPs adoption is to trust the AI to automate these decisions and actions. This is key to be able to manage more complex systems and reduce the workload on your IT Ops teams. IBM approach to AI based automation is to always ensure explainable AI, which allows human users to comprehend and trust the results and output created by machine learning algorithms.</li></ul><p>3. <strong>What kind of skills are needed to implement a strong AIOps approach?</strong></p><ul><li>From an implementation perspective most traditional IT administrators should have all the necessary skills to build out and manage an AIOPs solution. In terms of the AIOPs end-users, IBM’s philosophy to AIOPs is that the more people that are able to have their hands on an AIOPs solution, the more visibility and insights it can provide to different levels of your business, the more valuable it becomes. That is why we have designed and built our AIOPs solution to be relevant for SREs, developers, managers and IT administrators, and even business executives.</li></ul><p><strong>4. What is the average adoption time before tangible results can be achieved?</strong></p><ul><li>Certain aspects of AIOPs are designed to start providing value from Day 1. These are solutions in the observability and application resource management space. The insights provided by these tools are immediately valuable. The AI part of AIOPs requires some additional time to collect “good data” to learn about your environments and workloads. In the event management and anomaly detection spaces, most customers are able to start seeing useful results within a few weeks which continue to improve with time.</li></ul><p><strong>5. Is AIOps a concern only for our IT/monitoring teams or can our application teams and lines of business realize value from it as well?</strong></p><ul><li>AIOps done right is of great value to your LOBs, since the ultimate goal of AIOps is to keep applications running in a performant manner, avoiding or minimizing problems, as cost effectively as possible, all while maintaining compliance. Although many of the folks that will be implementing and using the AIOPs solution will be part of the IT organization, AIOPs provides value throughout your business. A comprehensive AIOPs solution will provide visibility and critical insights, as well as automation which delivers value across the business, application, and IT organizations.</li></ul><p><strong>6. How widely do we have to implement AIOPs for it to be effective? Do we have to coordinate across multiple teams and systems simultaneously to execute a working solution?</strong></p><ul><li>A key part of AIOps is using AI to automate configurations and processes that previously required manual configuration and maintenance. IBM’s approach to our AIOPs tools is to derive value from them on Day 1, via auto-discovery, auto-instrumentation, and agentless communication, wherever possible. AIOPs tools should also be able to automatically build and understand the dependencies and relationships between your applications, services, and hardware resources. Building this context in an automated and dynamic way can save countless weeks of manual mapping and continued maintenance.</li><li>An ideal AIOPs solution should be deployed across all systems, but this is by no means required. We see many of our customers start with a much smaller footprint and then expand as they develop and mature their AIOPs processes. However, it is often very useful to apply an AIOPs solution throughout the full stack of a key application, from a mobile app or website, to back-end VMs and containers, databases, network infrastructure, storage, etc. This is the most effective way to understand how to resolve issues that affect the health of your customer-facing business apps. As highlighted above, AIOPs tooling can help to tie together the entire stack and apply (previously vetted) automation to resolve issues more quickly and without human intervention.</li></ul><p><strong>7. How much data and time is required to effectively start getting value from AIOPs and automated decisioning?</strong></p><ul><li>In the observability and application resource management spaces of AIOPs, there is not any historical data needed to start decisioning. What is important is to have access to production workloads or data that closely mimics your real world usage (whether they be in test or production environments) so the tools can provide relevant insights and recommendations for your business within hours of implementation.</li><li>In the anomaly detection space, high quality input is an important factor in ensuring high quality outputs. We recommend starting with a smaller number of key performance metrics (KPIs) and then adding more data over time, if needed to refine the results. We have found that this approach achieves better results than ingesting all the available data and trying to make sense of it. Usually, we recommend at least two weeks of real world data to start seeing relevant outputs.</li></ul><p><strong>8. Are there industry standards in the AIOPs space to integrate different systems?</strong></p><ul><li>Rest API’s have become much improved over the years as a universal way to interact with services, and there are some standards in areas like network monitoring to collect device data. There are also new APIs and SDKs in software development such as Open Telemetry to standardize the instrumentation, collection, and generation of metrics, logs, and trace data to analyze your applications performance and behavior. However, for the most part there is no one accepted method across the industry for operations monitoring standardization. IBM’s approach has been to use standard integrations wherever possible, but to also be able to ingest data from traditionally unstructured sources and use AI to analyze and correlate it. This allows for insights from valuable data sources without additional manual intervention.</li></ul><p><strong>9. Do I need to have skilled resources in data science to use Watson AIOPs.</strong></p><ul><li>No. IBM has already provided the algorithms that can train on your data to build it’s unsupervised models. These models are designed to learn from user feedback and continuously update themselves when new data is available.</li></ul><p><strong>10. Do I have to write and maintain my own automation to integrate with your AIOPs solutions?</strong></p><ul><li>This is always an option, and in some cases necessary, however, IBM’s point of view is that our automation should start providing your value right out of the box. So we have offerings that automatically perform many of the tasks that were previously manually done, such as agent configuration and updates, or right-sizing application resources, and many more, all while maintaining your governance and audit policies.</li></ul><p><strong>11. What is the difference between Watson AIOPs and WatsonX?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vNsy8TfTrz26qHILWWNJWw.png" /></figure><ul><li>WatsonX is IBM’s latest AI offering and is an entire AI platform that is comprised of three components watsonx.ai, watsonx.data, and watsonx.governance. Each of these components help you to utilize foundation models, generative AI, machine learning to adopt AI into various aspects of your business, while maintaining transparency and explainability of your data and meeting your compliance and regulatory requirements.</li><li>IBM’s Watson AIOPs offerings are built for AI powered IT operations and not as a general purpose AI platform. Check out these links to learn a lot more:</li><li><a href="https://www.ibm.com/watsonx">https://www.ibm.com/watsonx</a></li><li><a href="https://www.ibm.com/products/cloud-pak-for-aiops">https://www.ibm.com/products/cloud-pak-for-aiops</a></li></ul><p>I’m sure there are many other questions you may have but I hope this gives you a better understanding of how AIOPs can bring value to your organization.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=78a951b92ad2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Observing my Java App on Liberty with Instana]]></title>
            <link>https://medium.com/@kmukherjee/observing-my-java-app-on-liberty-with-instana-3f47e3eb3873?source=rss-d1b59b115e6d------2</link>
            <guid isPermaLink="false">https://medium.com/p/3f47e3eb3873</guid>
            <dc:creator><![CDATA[Kaushik Mukherjee]]></dc:creator>
            <pubDate>Wed, 28 Dec 2022 16:47:40 GMT</pubDate>
            <atom:updated>2026-04-07T18:06:59.242Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*NUjkCBBLnuxjT_Tnisft4A.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*AyucUfa4YoSUJM3cBGa5Bg.jpeg" /></figure><p>I’ve been playing around with Instana for observability quite a bit lately and also trying out some of the latest Java development tools with Open Liberty. Others may be interested in setting up an environment like this so I documented my steps and what the experience was like.</p><p>I come from a Java development background, and one of the most difficult steps of getting a new team member started was setting up their development environment to be identical with the rest of the team. Things have gotten a lot easier in recent years and I’ll show you all the tools I used in my environment to simplify the process.</p><p>My goal for this project was to develop a small REST based application using Open Liberty, run it in a Windows Subsystem for Linux environment (WSL), and then monitor the golden signals for the application using Instana, and pinpoint any problem areas.</p><p>To get started I set up VSCode and installed the Liberty Tools extension. I also have a number of other extensions installed, like the Java tools extension and Maven extension but they aren’t necessary for this project, but you can install them if you like and you plan on doing more Java development in the future. You’ll also need to make sure that you have a Java SDK (I would recommend Java 11 or higher), Maven, and Git as well. Here’s where I downloaded everything:</p><p><a href="https://developer.ibm.com/languages/java/semeru-runtimes/downloads/">OpenJ9 Java</a></p><p><a href="https://maven.apache.org/download.cgi">Maven download</a></p><p><a href="https://git-scm.com/book/en/v2/Getting-Started-Installing-Git">Git Install</a></p><p>If you’d like to have also like to develop and run your code in WSL, you’ll need to install the WSL and Remote Development Extension Pack from Microsoft which includes the WSL extension. If you are on Windows these extensions allow you to open a remote folder or container on a remote machine or WSL.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EzvnJz9scoWWWslCiMaJhQ.jpeg" /></figure><p>You may also need to update your Linux distribution so that your VSCode server can start up in your remote environment.</p><p>From here all you need to do is clone one of the projects from the OpenLiberty guides on GitHub.</p><pre>git clone https://github.com/OpenLiberty/guide-rest-intro.git</pre><p>If you want a more detailed walkthrough of this project you can run through the <a href="https://openliberty.io/guides/rest-intro.html">“Creating a Restful Web Service”</a> guide on the OpenLiberty page.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1qHnJgpbvYFvcdyxlbVApA.jpeg" /></figure><p>Now you can use <a href="https://code.visualstudio.com/download">VSCode</a> to open up a workspace and if you’d like you can start modifying your code (not required). The easiest way to do this is to cd to your work directory and start a VSCode instance from there.</p><pre>cd ~/guide-rest-intro<br>code .</pre><p>One of really cool features of Liberty is its first-class integration with Maven. This means that I don’t have to download or install an OpenLiberty server. The maven plugin takes care of all that for me and I can just work on my code, and I can be sure that each Liberty server instance is isolated for each of my projects that I may be working on. The plugin also monitors any code changes in “dev mode” and hot reloads and deploys any code changes on the fly, greatly speeding up your development, test, debug loop.</p><p>The Liberty Maven plugin is already included as part of this project, but if you have another Liberty project you are working on you can simply add the following lines to your pom.xml under your &lt;plugins&gt;:</p><pre>&lt;! - Enable liberty-maven plugin →<br>&lt;plugin&gt;<br>    &lt;groupId&gt;io.openliberty.tools&lt;/groupId&gt;<br>    &lt;artifactId&gt;liberty-maven-plugin&lt;/artifactId&gt;<br>    &lt;version&gt;3.5.1&lt;/version&gt;<br>&lt;/plugin&gt;</pre><p>Once you get to this point you can check that you have everything configured correctly by starting up your application. You can build and start the application right from the “Liberty Dashboard” but right-clicking on the first “guide-rest-intro” project and clicking “<strong>Start</strong>”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*owBjhkSwFEGKw5ODO97tKA.jpeg" /></figure><p>Check that you Liberty server is up and running:</p><p><a href="http://localhost:9080/">http://localhost:9080/</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_qJ6-I5Psa5fim3FTPXPBw.jpeg" /></figure><p>Next, make sure your application is working by executing a HTTP GET request through your browser to display all the system properties</p><p><a href="http://localhost:9080/LibertyProject/system/properties">http://localhost:9080/LibertyProject/system/properties</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aPZZbZwLV-YoqRF0WpLFGg.jpeg" /></figure><p>Now that we have our application up and running let’s start monitoring our environment with Instana. I already have a shared Instana backend installed so now I need to install an Instana agent on my Linux host that is running my application.</p><p>On the left panel, click on the three dots towards the bottom and then click on “Agents”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XmYOPRqsoOJboEoBfinVvw.jpeg" /></figure><p>Next click on the “Install Instana Agents” button at the top of the page.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E8YdcbWKjY3P6m5HSv6ILA.jpeg" /></figure><p>From this page we can choose the type of agent we want to install. In our case we need a Linux agent. Select Linux on the left side under OS and then check the boxes for Dynamic for Agent Mode, Eclipse OpenJ9 11 for Agent Runtime, and Silent for Installation Mode. This will give you a “one-liner” to download and install the agent. Notice that the necessary API’s keys are also part of the command. I have blocked those out in my screenshot below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KZfZXQVhgoXZvlLbsOLmZg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4PmYwCQkz-A3-XV1ZZcdMQ.jpeg" /></figure><p>Once the agent install is complete, I started the agent issuing the start command:</p><pre>/opt/instana/agent/bin/sudo ./start</pre><p>Check the status of your agent by running the status command:</p><pre>/opt/instana/agent/bin/status<br>The amount of memory to be used by the agent seems too large (9007199254740989 KB, which is bigger or equal to the 15309864 KB total system memory), so the agent may actually not be restricted to a fixed amount of memory. To avoid consuming too much memory, the agent will use up to 512m.<br>If you want to make available more or less memory to the agent, set the &#39;MemoryMax&#39; property accordingly in the &#39;instana-agent.service&#39; SystemD unit.<br>Checking Host machine for TLS 1.3 support<br>IBM Semeru Runtime Open Edition 11.0.17.0 (build 11.0.17+8)<br>Running ... (pid 1509)<br>kaushik@KTop-WSL:/opt/instana/agent/bin$</pre><p>You can also run through the agent logs in <em>/opt/instana/agent/data/log/agent.log</em> to make sure you don’t see any connectivity errors.</p><p>Now it should only take a few minutes before the new agent shows up as another host in the Instana console, and starts reporting metrics. If you look closely at the Instana logs you will also see a number of sensors getting deployed for automatically detected technologies. For example, when Instana detects the Liberty server you will see this in the logs:</p><pre>2022–12–27T12:09:06.846–05:00 | INFO | om.instana.plugin.javatrace-5187 | JavaTrace | com.instana.sensor-java-trace - 1.2.428 | Connection with JVM with PID 5187 on 127.0.0.1:46465 established<br>2022–12–27T12:09:16.416–05:00 | INFO | stana-agent-scheduler-thread-5–2 | turesManagerImpl | com.instana.agent-bootstrap - 1.2.25 | Installed instana-websphere-liberty-sensor<br>2022–12–27T12:09:16.436–05:00 | INFO | 4f5–0889–491b-9cb0–0925adb5b5a8) | WebSphereLiberty | com.instana.sensor-websphere-liberty - 1.1.2 | Activated Sensor for JVM 5187<br>2022–12–27T12:09:16.446–05:00 | INFO | ugin.websphereliberty.5187attach | WebSphereLiberty | com.instana.sensor-websphere-liberty - 1.1.2 | Attaching to JVM with PID 5187<br>2022–12–27T12:09:16.585–05:00 | INFO | ana.plugin.websphereliberty-5187 | WebSphereLiberty | com.instana.sensor-websphere-liberty - 1.1.2 | Accepting data from JVM with PID 5187 on 127.0.0.1:43375<br>2022–12–27T12:09:16.592–05:00 | INFO | ana.plugin.websphereliberty-5187 | WebSphereLiberty | com.instana.sensor-websphere-liberty - 1.1.2 | Connection with JVM with PID 5187 on 127.0.0.1:43375 established<br>2022–12–27T12:09:41.827–05:00 | INFO | instana-http-client-thread-1 | LoggerEndpoint | com.instana.agent - 1.1.669 | JVM (5187) - Spent 34290 ms and 0 KiB for instrumentation [trace] (Used: 0 KiB, Max: unlimited) (JVM uptime: 49075 ms)</pre><p>Let’s go to the main Instana page and navigate to the relevant dashboards for our application. We can start by taking a look at the Infrastructure console to see all the hosts that are currently being monitored. Click on “Infrastructure” on the left side:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1FW6kM59tmeDM1fm0ARwxQ.jpeg" /></figure><p>If you have a lot of hosts already added to your Instana environment, you may want to filter your results. In my case I used hostname to narrow my results:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wdvToHkL2-HYTlk0-UO1nw.jpeg" /></figure><p>Each box in the tower above represents a different process that is being monitored on my host. If you hover your mouse over the different boxes in the tower you can see more details about each process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nVWHlkOliqoDJlnYqcU5yw.jpeg" /></figure><p>You can also click on the “Comparison Table” tab to get a list view of all the monitored hosts.</p><p>On the page above we can then click on “Open Dashboard” and we see a much more detailed view of my Linux host with metrics such as CPU, memory, and load. You can even see the metrics of each individual CPU on the host.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*25zSHmZnDI3seq3nXpNwyA.jpeg" /></figure><p>If you haven’t already, now would be a good time to generate some load against your REST application so that we have some traffic to analyze. I used Liberty’s build in REST API tool based on Swagger/OpenAPI, but you can use any tool you’d like such as Postman or JMeter. If you’d like to use the same steps as I did simply add the <strong>mpOpenAPI-3.0 </strong>to your Liberty server.xml</p><pre>  &lt;featureManager&gt;<br>      &lt;feature&gt;restfulWS-3.0&lt;/feature&gt;<br>      &lt;feature&gt;jsonb-2.0&lt;/feature&gt;<br>      &lt;feature&gt;mpOpenAPI-3.0&lt;/feature&gt;<br>      &lt;feature&gt;monitor-1.0&lt;/feature&gt;<br>  &lt;/featureManager&gt;</pre><p>You can then view all your endpoints at this URL and execute requests directly from the web based user interface. For my application I added some new GET and POST methods using JAX-RS to the cloned application for some other demos I’ve been working on, but you can skip that step for the time being unless your are already familiar with the JAX-RS programming model. In the original REST project you will only see the /system/properties endpoint:</p><p><a href="http://localhost:9080/openapi/ui/">http://localhost:9080/openapi/ui/</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Sahr6ol-81U4yVKvAWD4Xw.jpeg" /></figure><p>From here let’s see what kind of information we can gather about our REST services that are running on our Liberty server. Click on “WebSphere defaultServer” to jump to the dashboard for our Liberty Server</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1CEHD1Ngen_KUeihwR81ZA.jpeg" /></figure><p>Here we can see more details specific to our Liberty server and the services running on it. We can then click on the link under the Services and we can see all the golden signals associated with all the externally exposed endpoints.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rl4IV_c2BSRpvypG5RGPeA.jpeg" /></figure><p>From this dashboard it looks like all of our calls are successful, however, we seem to have some spikes in latency. If we click on the Analyze Calls button at the top of the page we can see each individual call that was made against our application. Instana doesn’t sample it’s trace data, so we can see the details of every call without losing any important information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8pxwE0NJlAh_1nzVTYJ5Pw.jpeg" /></figure><p>From the dashboard above we can see that we have 30 GET requests and 15 POST requests. However, the latency of the GET requests seems a bit higher than we expected. We can now click on the GET requests to see each request:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sq7Yn6dOrGw8fxzhFaOwuQ.jpeg" /></figure><p>As we drill into this list of GET calls we can ee that most of our GET calls have a latency between 1 and 3 milliseconds, which is reasonable. However, we have one endpoint which is taking well over 2 seconds to respond. In this example I created an endpoint which reaches out to an external service that happens to sit on a very slow database. From the information we have gathered we can now work on improving the latency of the /<em>helloslow </em>endpoint. We may want to issue a faster timeout if the response takes too long, add some caching, or if we have access to the external database we can investigate if we need to change it’s storage type or add more resources.</p><p>Keep in mind that everything we were able to view and analyze so far was done without any configuration or building custom dashboards. This article should give you a good idea of how to get a Java application up and running quickly and then observing your services. From there you can explore all the other powerful capabilities built into Instana. If you’d like to try out <a href="https://openliberty.io/">Open Liberty</a> or <a href="https://www.instana.com/">Instana </a>check out the links below. The <a href="https://www.instana.com/apm-observability-sandbox/">Instana APM Observability Sandbox</a> is a great way to get familiar with Instana and see if it makes sense to add it to your observability toolbox.</p><ul><li><a href="https://openliberty.io/">Open Liberty</a></li><li><a href="https://www.instana.com/apm-observability-sandbox/">Instana - Play With Instana Login</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3f47e3eb3873" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>