Load Impact – a leading SaaS performance testing provider based in Sweden – added a new capability that gives developers a more complete picture of their performance tests. Most cloud-based load testing tools only measure web application responses, but the new Server Metrics agent from Load Impact provide insight into what’s happening on the target servers themselves.
Cloud-based load testing tools have grown in popularity as developers build and test geographically-dispersed, high-traffic public web applications. However, while this new generation of tools offer easy setup and sophisticated traffic simulation, they lack the server-side component that exists in traditional on-premises tools. A tester can only see that website performance is degrading, but can’t tell why. In a press release this week, Load Impact described how their new Server Metrics components fills the gap.
The new Load Impact Server Metrics Agent provides equally valuable insight but from the server-side perspective, so developers can more easily identify where performance bottlenecks and scalability problems exist in their code or infrastructure, and operations professionals can better allocate resources.
…
Server metrics are a pretty standard part of old-school load testing software, which is complex, inflexible, super expensive and requires hardware and specialists to manage and use. The Load Impact Server Metrics agent is really the first of its kind. None of the other new-generation cloud-based load testing services have it today.
Developers could solve this issue today by cobbling together a solution with any cloud-based load testing tool and a server-side agent provided by a monitoring tool such as New Relic, but Load Impact sees the value in a single solution that natively integrates the entire performance data set.
By installing the Server Metrics Agent on one or more target systems, our load testing server can pick up some internal measurements during the test and add those to the same data set. Load Impact supports collecting data from up to three different target machines during a test, so it’s possible to get internal measurements from a fairly complex setup as well. The advantage of this is quite obvious. Even if it would be possible to log this data separately on the target machines, you would end up with the task of trying to synchronize the time stamps of the internally generated data series with the data from Load Impact.
Load Impact CEO Ragnar Lönn spoke to InfoQ and explained that this toolset was born from their experience as performance management consultants. He says that developers are moving towards always-on, SaaS-based load testing tools, but they needed a way to see what was going on inside the target server environment. Lönn sees customers doing real-time adjustments to server environments while load tests are running, and now those customers have a much better idea of which server-side resources to modify to improve performance. A blog post about this release demonstrates that server-side data is captured and displayed to users in real-time.
The Server Metrics Agent captures basic server-side metrics such as CPU, memory, disk space, and network consumption. To get additional metrics, Load Impact designed their Agent to accept plugins compatible with Nagios, a standard tool for infrastructure monitoring. The Nagios Exchange includes thousands of plugins that are capable of things such as monitoring database client connections, inspecting individual Windows processes, and reporting on swap frequency on a Linux server. This Agent can be installed on servers running Debian-based Linux distributions or Windows Server 2008R2 / 2012.