I recently read a TechTarget article by Stephen Bigelow – “Look beyond APM to unified performance monitoring.” In this article, Mr. Bigelow explains that traditional performance monitoring tools miss the market because tools often serve one or two silos in the organization. As a result, businesses rely on multiple tools to monitor for performance issues across multiple silos (application, storage, virtualization, server, etc.).
Often it is more the failure of the IT organization that creates tool sprawl than the tool vendors not providing unified performance monitoring. If there is not a dedicated monitoring team across the organization, defining monitoring objectives (and tools) falls to individual admins. This is evidenced by the fact that you will see multiple implementations of a single tool in one company. Without a dedicated monitoring team, there is a lack of knowledge to understand what tools are already owned by the company and how they can be used.
A dedicated monitoring team also provides the big picture view into how a service should be monitored to mitigate gaps in visibility – from the hardware to the app to the storage and database and everything in between. Often, dedicated monitoring teams will decide to standardize on a suite of monitoring tools because the suite shares a common UI, alerting mechanism, database, naming conventions, etc.. When there is a gap in functionality, then the team will look for another tool to fill that gap.
There are suites of tools that offer performance monitoring for multiple silos on the market today. However, many of these traditional suites don’t provide the contextual visibility into how the entire app stack is performing because these tools are generally written by different development teams who are focused on a particular silo (storage, virt, etc.).
Today, there are many new vendors that develop performance monitoring tools that provide capabilities for the SME they are serving, but also consider the application’s perspective. Some examples of this approach include application aware network performance monitoring (for the network admin), transaction tracing (for developers or application admins), application aware database performance analysis (for DBAs and devs), and app aware infrastructure performance monitoring (for IT Operations or system administrators). For instance, app aware database performance analysis allows you to filter response time for a particular application, and then see the top queries and waits (wait types / wait events) for that specific program. App aware infrastructure performance monitoring helped one customer diagnose a perceived application performance issue to find that it was actually a storage I/O spike caused by a daily backup procedure.
Monitoring vendors are getting better at bridging the gap between monitoring silos with contextual linkages, but it is still up to the IT organization to make a conscious decision to minimize tool sprawl.