Premature Evacuation?

cloud for blog
Watch this on-demand webinar talking about APM strategies for rapidly changing data center landscapes

For those who follow VMware, you know the company has been systematically exiting markets and product lines for several quarters now. This particular departure was announced early this May:  End of Availability of vFabric Application Performance Manager (vFabric APM)

The history of VMware’s APM strategy reads like a Greek tragedy (the tale of Icarus comes to mind). VMware’s foray into APM began way back in 2008 with the acquisition of an Israeli company – B-hive Networks – for a 20x+ revenue multiple. The company had some cool technology for measuring hop-by-hop response times in distributed web applications by tapping into the vSwitch network traffic.

The technology was pulled from the market (and some say rebuilt entirely) and then launched as VMware AppSpeed in 2009. Version 1.5 was launched amidst major hoopla around new hooks to Dynamic Resource Schedule (DRS). Full of promise, the idea was to dynamically re-configure a private cloud based on server-to-server response times to optimize application delivery.

In between, VMware bought Hyperic and SpringSource and out of this marriage was born a Java bytecode instrumentation solution – vFabric AppInsight. Hyperic, AppInsight and AppSpeed got bundled up into a single SKU – vFabric APM. Hyperic provided resource utilization stats from the virtualized infrastructure. AppInsight provided a deep dive into apps built on SpringSource. AppSpeed provided response time information. All of a sudden, VMware was only one dimension short of a Gartner full house in APM.

AppSpeed got the knife in January of 2012. AppInsight slid off the VMware price list on June 1st 2013. And with those two strokes, VMware no longer has an APM offering. All licenses convert to the Hyperic resource utilization monitoring software.

So what does this mean?  Well, monitoring application performance in virtualized environments is not only an unsolved problem…it’s now a wide open market looking for a committed APM leader. It’s surprising that VMware ceded a $2.2B market that is growing at 13% a year (Source: IDC “Worldwide Application Performance Management Software 2012–2016 Forecast” 2012), but let’s not armchair quarterback that decision here…

Instead, let’s take a look at three primary challenges when it comes to monitoring application performance in a virtualized environment.

Challenge #1: Finding a stable, reliable, and comprehensive place to capture data that you can trust about application behavior.

There are issues with all the different instrumentation and data collection approaches, be them network-based, host-based, or application-based. VMware never quite licked this problem, and high-profile cases like the New Relic/Heroku/Rap Genius example highlight why it’s so important.

Challenge #2: Tying together transactions that may span multiple platforms, operating systems, and programming languages

Many of the most important enterprise apps still run on a mixture of virtualized and physical servers and across multiple platforms. VMware never had much incentive to figure out the physical side of the equation.

Challenge #3: Processing the data fast enough to provide useful information

VMware used to articulate a “Monitor, Respond, Alert” vision, where infrastructure could react in real-time (and more importantly, faster than a human) to adjust to meet changing resource demands. However, none of their APM offerings were constructed with a particular emphasis on performance, throughput or low latency.

Why the premature evacuation, we may never know the real answer to this question.  Perhaps VMware’s constantly shifting strategy created confusion for their customers that slowed adoption of much-needed APM software in these environments. Much like IBM or Cisco can stall markets and starve start-ups with a press release, VMware seemed to keep a lot of customers on the sidelines with a compelling vision, but lackluster execution at the product level.  Whatever the reason, the silver lining is more market opportunity for smart vendors that really want to focus on APM to solve these non-trivial challenges and explore fresh approaches to application performance management for distributed application environments.

If you are interested in learning more about why the adoption of cloud platforms and infrastructures, virtualization and Agile application development methodologies requires a fresh approach to APM, download this whitepaper titled, “Why Gartner’s 5-Dimensions of APM Are Increasingly Irrelevant to You (And What You Should Do About It)”.