APMdigest recently published a list of the top factors that impact application performance. Answers varied, but a common theme was that application complexity has become a major roadblock to high performance. EMA Analyst Julie Craig put it perfectly when she described an environment where the next problem could come from anywhere — not just from the application code.
In this environment, Server Administrators face a huge challenge. They’re responsible for supporting the backbone of these complex applications, but the vast majority of these IT Operations team members have no knowledge of how a specific server fits within the production business applications. In many cases, they don’t even know which servers communicate directly with each other.
At BlueStripe, we recently interviewed server admins about the problems they face when dealing with application performance issues. The majority talked about a lack of information at the application level, with many saying that they did not even know which applications ran on which servers (and vice versa).
Here’s why this is a problem: more than half of all the companies surveyed in our recent IT Operations survey indicated that they rely on bridge calls as the preferred process for application troubleshooting. These calls could be much less frustrating (for Server Admins personally) and far more productive (for their employers) if server admins could see how the systems they manage relate to the applications with the performance problems.
Time to Switch the Point of View
It’s not the server admins’ fault that they don’t have this information. Server monitoring tools just haven't kept up with changes in application environments. For individual server admins, performance monitoring hasn’t grown much past Task Manager and Performance Monitor. Their tools provide commodity machine performance metrics (CPU, memory, disk I/O), but they can’t show which applications run on those machines, or anything about response times, or how individual applications connect to other machines within the infrastructure.
This creates a disconnect within many IT Operations teams where domain experts (Unix, Java, Database, etc.) have their own tools to determine how individual systems are operating, but there’s no larger picture of the application as a whole. The problem solving process is destined to fail before the first problem ever shows up. The team is responsible for delivering business services, but their tools don’t take those services into account when judging individual system performance.
Finding a Solution
To address the disconnect between what server admins can easily see and the requirements of today’s distributed applications, IT Operations teams need to look for technologies that provide an application-centric view of what’s running on a specific server and what client and back-ends are connecting to each application.
Put another way, server admins need to see which URLs and requests are slow, how CPU, memory and disk performance are impacted by each application, and measure response times for databases and other back-end dependencies. They need the ability to look at historical data, so that users can review what happened leading up to and through a performance issue, instead of just looking at what’s happening in the moment.
With this increased visibility into the intersection between applications and infrastructure, server admins can play a greater role in contributing to the IT Operations team’s primary goal of delivering application services.
Vic Nyman is Co-Founder and COO of BlueStripe Software.