In my last post, I’d mentioned that we had delivered a Splunk add-on for IBM WebSphere Application Server environments on SplunkBase. I am proud to announce that it’s a full fledged app now, complete with some very cool views, dashboards and saved searches.
When we saw Splunk being used across a huge number of customers for the primary purpose of distributed application management, and realized there were things they were all doing over and over again – we thought, wouldn’t it be cool to simply make an app available for those things? So in many ways, the Splunk for IBM WebSphere app embodies the things that most of our customers currently do or would want to do in WebSphere Application server environments. With the Splunk app, we not only collect data important for IBM WAS (log files, metrics, configurations), we also package in views that use the information to do things like performance monitoring, security or configuration change tracking or troubleshooting.
I do want to put in a couple of notes – 1. While enabling PMI ( which is what you need to do to let Splunk collect metrics, only collect the metrics you are really interested in: jvm memory usage, database connections, transaction times etc – collecting all will result in some overhead). 2.We’ve created a number of views and dashboards with the idea that they are really templates or starting points. Your custom environment may need a different view, cut and sliced differently from how we’ve done it. So, we are making no claims to be perfectJ and since Splunk will also pull any of your custom application logs in the WAS log directories, you are welcome to use our templates to create specific, custom views for your environment.
What’s cool about the views?
Well, first of all, there are many of them! They are categorized into performance, security, configuration tracking and troubleshooting views: if there are some views missing in your mind for these specific areas, do let us know by emailing me directly(email at the bottom). The idea is to be able to easily correlate performance or availability related events to exceptions/errors logged by a JVM or a recent configuration change in the environment.
Performance views include visibility into thread pool sizes, servlet response metrics, transaction response times, JVM memory usage, garbage collection times etc…your pick of typical hotspots in Java application environments. Configuration tracking let’s you view what’s changed in your environment recently. Security views collect all the security related errors and warnings in a single view over time. And troubleshooting views simply show you recent activities that may have caused errors in your environment.
Take an example situation: your application has sporadic connectivity to the database – your logs show a ConnectionWaitTimeout exception occurring around the times that connectivity was interrupted. You then look at your database connection pools chart and notice that your PercentUsed is consistently high. This means some tuning is required to get your database connection pool size to an optimal level.
We suspect that ANY java environment (not just WebSphere) would have similar looking views (partly because we’ve seen some of these at our customers and partly because we hear sufficient griping about things like garbage collection and message queues). So even if you don’t have a WebSphere environment, take a look at the screenshots here http://bit.ly/be1c9p and tell us what else you would like to see. My email is at the bottom of this post
What’s next for us?
This app is a first step for us. Since Splunk is used very broadly to troubleshoot and monitor applications across every type of industry, our goal continues to be to make it easier for our customers to get value from the power of Splunk. We continue to evaluate a possible .NET app. If you’re interested and want to participate in defining requirements for it, email us: ljoshi AT splunk.com and jgrant AT splunk.com