AppDynamics Prepares For The Cloud Era

AppDynamics, the San Francisco based Java Application Performance Management vendor, today announced the release of AppDynamics 3.0, designed specifically for Java applications in the cloud, virtual and hybrid environments. Application performance monitoring is crucial to any organization with mission critical applications. AppDynamics is a startup trying to compete in the space dominated by CA, IBM, etc.. With this new release, they are jumping far ahead from some of the solutions that exists in the market.

The Problem

Whether it is the traditional computing era or the current cloud based era, Application Performance Management (APM) is one of the key focus areas of IT. The first generation APM solutions were tailored towards the demand of monolithic applications running on top of the traditional IT architectures. The last 10+ years saw the emergence of an increasingly distributed architecture with SOA, Virtualization and, now, cloud. Existing APM solutions are not well prepared for this kind of highly distributed ecosystems. The elastic nature of cloud imposes additional problems in terms of a need for automatic discovery of provisioning and de-provisioning of cloud nodes and the ability of the APM agents to automatically follow the transaction flow and end user sessions across these additional (or lesser) nodes. Being able to monitor the performance issues in an application present in such a dynamic environment is a difficult problem. Moreover, being able to do it without any additional overheads is even more difficult. Now if we add the agile development philosophy of modern applications into the mix, we have a really messy problem to solve.

AppDynamics Solution

AppDynamics uses two patent pending technologies, Dynamics Flow Mapping and Deep On-Demand Diagnostics, to solve this problem. Dynamic Flow Mapping technology automatically detects and tracks the application services and tiers, the business transactions, and the end-to-end flow of each business transaction across multiple services. The low overhead agents present at the middleware layer uses automatic Byte code injection to talk to the application and get diagnostic data. Deep-on-Demand Diagnostics technology enables deep visibility automatically as needed by offering in-flight response analysis offering early warning, pattern performance analysis to identify bad performance and let the diagnostics go deeper to check the root cause and call stack analysis which offers deep visibility at the code level. The version 3 of AppDynamics thus offers Application Performance Management solution for Java applications running across thousands of nodes in the cloud.

Another issue with the traditional APM solutions is that the management console doesn’t scale well with the scalability available in the cloud. Typically, one management console supports 200 nodes. If an organization runs their application on top of thousands of nodes, the IT will have trouble managing 5+ management consoles trying to identify the factors affecting the performance. This adds too much complexity to the management of application performance. AppDynamics scales well with the cloud where the agents are deployed and supports up to 1000 nodes with one management console.

How much does it cost?

They offer both a free version and paid version. Their free version, called AppDynamics Lite, offers most of the features available in the first generation APM solutions. They are using this free version to disrupt the market and gain some market share against the likes of CA, IBM, etc.. Their paid version, called AppDynamics Pro, offers additional features including the support for cloud based architectures. Today’s release, AppDynamics 3.0, caters to the paid users.

What is new in this release?

This release of AppDynamics is tailored to suit the cloud environment better. It introduces new features like

Dynamic Cluster Aggregation & Intelligent Instance Lifecycle Tracking which discovers the provisioning and de-provisioning of nodes in the cloud as they happen and self-learns “normal” performance levels accordingly
Support for up to 1000 cloud nodes to report to single AppDynamics controller

Additionally, AppDynamics 3.0 also supports automatic real-time memory leak detection and root cause diagnostics.

Final Thoughts

AppDynamics is in a face long dominated with big vendors and they have a really tough task in the market. But they are shaping up really well. Two of their customers (which they happily showcase) are Priceline and Netflix. Both these services are customer facing ones with high traffic numbers. They live on a distributed environment with too many moving parts. Monitoring the performance of these applications is not an easy task and AppDynamics is doing a great job with just 2% additional overhead at a fraction of the cost compared to the offerings from big vendors. They are on right track and it will be interesting to watch their evolution in the coming years.