Application Performance Management: Laying The Groundwork
Posted by Michael Biddick on July 26, 2010
This week I was working with a client, we will call him Jim, who was telling me about their mission critical business application. It's primarily a billing system, but also has service catalog and fairly detailed security and reporting modules built into the system. As a web-based application, the backend database is distributed and the application services just under 2,000 users. The CIO is seeking to establish key performance indicators around the application including performance metrics and SLAs. Unfortunately, there is currently no fault or performance monitoring in place. The current performance of the application is judged based on anecdotal calls to the help desk. Obviously, this will not cut it.
The community of APM vendors grows each week. As the industry finally adopts a service-based approach for IT management, the services really end up being applications at the end of the day. The tidal wave of ITIL process initiatives has been a big catalyst for APM projects as organization try and consolidate their legacy application and system management tools from the past 20 years and provide more proactive metrics around the critical business applications.
When we talk about end-to-end APM, usually vendors will come into environments where network monitoring and perhaps some system level monitoring already occurs. Jim's environment is very different. Discussing business dashboards, integration and real-metrics is a real problem for many organizations but does not resonate with Jim. Like a large percentage of the IT community Jim is still in the dark around application performance. For him, end-to-end APM is an almost unimaginable nirvana which seems out of reach and too complex for their environment. He has survived for years without it, so when the CIO asks him for metrics, he really does not know where to begin.
The bright side of my conversation with Jim is that many of the issues surrounding legacy APM applications will not come into play, the elegance of designing a solution in a Greenfield environment has real potential. The real trick is to select the right type of tool and architecture that can meet Jim's needs, scale as the organization grows, but also not break the bank (which for Jim is not very big). So, how does Jim get started selecting the right APM solution for the environment? Over the next few weeks we'll discuss our approach and provide a roadmap for anyone implementing APM. From building the business case, developing your architecture and selecting the tool and implementation, we will map out the steps required to try and select the right tool for the job.
We also would like to hear about your recommendations and ideas for each of the stages, so feel free to share APM success (and failure) stories with us in the comments.












Comment by Greg on July 26, 2010 12:21 PM
The notion of having survived for years without APM, yet needing to choose the a tool that fits the need of a "service based" approach, is dead-on accurate. So is the requirement of not having to break the bank. As you say, a VERY large percentage of IT organizations are in this position.
There are some free tools for testing app performance, including in production--here's one.
Reply to this comment
Comment by Raja Mukerji on July 26, 2010 5:05 PM
Good post, Mike. Thanks for taking on this important topic. As applications become more distributed and abstracted, it has become increasingly more challenging to understand the most basic performance metrics. That's why most application issues often aren't identified until performance has already slowed to a crawl, or an outage has occurred (a little like when the Check Engine light in your car turns on as your car is breaking down!) And even for environments with network monitoring / APM tools in place, it's not always straightforward to diagnose issues and baseline application performance. This is exactly why we started ExtraHop, and established a new market category called Application Delivery Assurance. We believe that understanding application performance doesn't have to add another layer of complexity to your environment. Rather, we provide passive, real-time analysis from L2–L7 for proactive early warning and accelerated troubleshooting across network, web, database, and storage tiers. Check it out -- http://www.extrahop.com/products/. We hope we have the opportunity to contribute further to this discussion.
Reply to this comment
Comment by Manisha on August 4, 2010 6:43 AM
Implementing any of the existing APM tools not only requires capital investment bu ongoing operational expenses for administering and then interperting the data coming out of the tool. Check out this Application AWare infrastructure performance management service. THis service utilizes a unique tool(patent pending Intelligent APM) and provides not just identification of issues, but also analysis and L2 level root cause analysis.
This is the new APM as a managed service model, one step above SAAS.
http://www.appnomic.com/index.php/Application-Aware-Infrastructure-Performance-Management.html
If Jim just wants to a quick insight into performance of transactions on his appln within the DC, then he can sign up for this beta service www.myappmeter.com.
Reply to this comment