AppDynamics Pushes Application Performance Management to the Cloud

Yap Inc., a Web company that runs its automated voice-to-text transcription business on a cloud computing infrastructure, needed a way to monitor and manage just how long it takes to receive voice mails from customers and turn them into e-mails or text messages. So it turned to start-up AppDynamics, which this week unveiled its application performance management (APM) software built from the ground up to manage applications running in the cloud, virtual and service-oriented architecture (SOA) environments.

Founded in 2008 by Jyoti Bansal, AppDynamics is unveiling its AppDynamics 2.0 solution which is designed to help companies identify and resolve problems with applications that are running in highly distributed environments (this is the first version with general availability; AppDynamics is officially coming out of stealth mode with the the 2.0 version).

Before founding the company, Bansal, now AppDynamics' CEO, led design and architecture for several products at Wily Technologies, including Wily Introscope, an APM system used to detect and diagnose performance problems in composite and Web application environments. While it gave organizations a much-needed tool for the Java and .Net environments, the system was only able to provide visibility into applications running on a single Java Virtual Machine. There was no real visibility into distributed transactions. So Bansal got the idea to architect a next-generation APM solution. "For these cloud and virtual environments, you need wide and deep visibility," Bansal says. Moreover, the solution needs to be able to dynamically act, based on what the APM system is learning about transaction and application performance. That is a key part of application performance management. It has to be much more dynamic." As he writes on his company blog, Bansal wanted to give IT professionals a way to not only monitor application performance, but also to find and fix root-cause problems--to combine traditional monitoring with the ability to "orchestrate capacity in the cloud."

That is one of the cornerstones of AppDynamics 2.0--the ability to monitor cloud-deployed applications and intelligently provision capacity as needed. "As performance slows down because loads go up, companies should be able to scale up capacity. The solution needs to handle cloud-bursting, so you can get temporary capacity," says Bansal. "And when the spike goes away, then the solution can get rid of that capacity."

Policy-driven workflows with AppDynamics 2.0 can be triggered to automatically address capacity problems. The solution ships with a cloud connector to Amazon EC2 and a set or pre-defined workflow templates to automate application provisioning and configuration. Application Dynamics plans to add other cloud connectors and templates in future versions. AppDynamics 2.0 provides organizations with a visual map of the application topology, and the software can dynamically discover all the application tiers and back-end services. The solution constantly looks at the application topology and can automatically discover new code. "Application logic is split amongst a lot of servers in these SOA and cloud environments, and often there are upgrades and new releases of code daily or weekly," says Steve Roop, VP of marketing at AppDynamics. "When so much is changing on the development side, the IT guys are the ones that have to catch that pitch."AppDynamics 2.0 also measures business transactions to monitor whether the applications are meeting service levels. The AppDynamics' Transaction Flow Monitoring capability gives organizations an inside view as to how each transaction performs as it moves within a distributed environment. AppDynamics 2.0 also leverages self-learning algorithms that automatically and regularly create performance baselines to create historical performance patterns and then compare those to current performance, says Roop. That way, the solution can distinguish between patterns of poor performance and one-time anomalies. AppDynamics Deep on-Demand Diagnostics is designed to perform rapid, root-cause analysis with no more than two percent overhead, according to the company. The solution's Performance Pattern Analysis uses advanced heuristics to detect critical performance incidents and capture code-level diagnostics. In addition, the solution is able to detect common performance problems, including memory leaks, stalls, deadlocks, thread contention issues and database connection pool health.

Rick Terrell, VP of Engineering at Yap, says AppDynamics' solution is helping Yap ensure that it's meeting the service levels its customers expect. "One of our main challenges is that we compete with services that use human transcription services. So we have to compete on turnaround times, by turning around text faster, and on accuracy," he says. That turnaround time is determined by how much audio came in and how quickly it needs to be converted into text and delivered to the customer, and Yap's key market differentiator is the speed in which it can convert audio to text. The company's core automated transcription application is Web-based and runs in cloud environment.

In addition to being easy to install, Terrell says AppDynamics' solution is easy to modify so that it can monitor application performance as it relates to each individual customer's service level agreements. "It is pretty easy to make up your own key performance indicators, instead of having to use the packaged ones that you typically see in application management solutions," Terrell says. "And we can tie those performance indicators all the way up to top of our application, so we can measure against whatever service we've contracted for."

With AppDynamics' solution, Yap can keep track of how much time it takes to complete work. "If a customer is sending us a certain amount of audio, say 100,000 voice mails a day, we keep track of how much time is used up to convert that." The APM solution has already uncovered a few problems. In one instance, a customer had a traffic problem on their end, which created a huge backlog of audio. As the network issues cleared up, the customer sent all the audio in one lump. "They dumped a couple thousand voice mails on in the time span of about five seconds, so AppDynamics saw and alert us to the spike. We were able to handle it, contact the customer, and clear everything up."  Now available, AppDynamics 2.0 starts at $5,000. 

InformationWeek's Informed CIO series lays out 10 questions to ask before proceeding with data center automation.  Download the report here (registration required).