I last researched the topic of Application Performance Management (APM) back in 2013 with a report entitled Application Performance Management (APM) in the Age of Hybrid Cloud. Hybrid cloud was then, and still is, an important topic. One key finding garnered from that research, however, was the fact that the term "hybrid cloud" is defined differently by virtually every vendor and IT organization.
For vendors, "hybrid cloud" solutions seem to be most frequently defined as "whatever cloud-related products we're trying to sell." On the IT side, the term "hybrid cloud" is most often defined as "whatever types of cloud services we're trying to integrate." Regardless, while hybrid cloud is still a topic of lively discussion on multiple fronts, I decided to let the dust settle for a bit and turn my attention to other important topic areas in 2014 and 2015.
For 2014, the focus areas were management of public cloud and DevOps/Continuous Delivery. In 2015, the topics included API management and DevOps/Continuous Delivery (again — a lot of interest and business value here). However, another very interesting outcome came out of the process of doing deep dives into a variety of seemingly unrelated topic areas: As is often the case, findings in one topic area always seem to contain breadcrumbs that generate questions relating to adjacent spaces.
The API research, for example, uncovered the fact that transactions leveraging APIs are more often managed from the perspective of the API Gateway (45% of respondents) than with APM solutions (32%). In essence, the Gateway has become another monitoring silo, which IT organizations are utilizing in standalone mode to track transaction performance and availability.
So at a time when software is becoming increasingly business relevant, IT teams are, in too many cases, retreating to the silo monitoring techniques of the past to track and troubleshoot application performance. This may well be due to the fact that they lack access to APM solutions. Nevertheless, as is always the case with silo-based monitoring, the problem is that monitoring the gateway alone results in too many gaps in visibility to efficiently automate end-to-end troubleshooting and root cause analysis.
The DevOps and Continuous Delivery studies uncovered APM-related breadcrumbs as well. The 2015 research, for example, found that while Continuous Delivery has a proven upside to the business, it is also siphoning both Dev and Ops resources away from the development and delivery processes and into production support.
Specifically, companies in which "Continuous Delivery" frequency increased by 10% or more during the prior year were 2.5 times more likely to experience double-digit revenue growth than their less nimble competitors. In other good news for the business, almost 50% of survey respondents reported that the increase in delivery frequency resulted in "higher levels of customer satisfaction."
At the same time and on the opposite end of the spectrum, the impact on IT is not as rosy. Approximately 50% of respondents reported that development is being drawn into the troubleshooting process more often; a similar percentage reported that operations is spending more time on production support as well. The culprit seems to be the increased production change volumes introduced by accelerated Agile and Continuous Delivery practices.
Survey respondents also point to "manual troubleshooting processes arising from production changes" as the #1 bottleneck slowing down the Continuous Delivery pipeline. So while acceleration of the Continuous Delivery process has a strong impact on the business bottom line, increased time spent on production support is reducing the time Dev and Ops teams can actually spend rolling out new services.
These findings point to a need for "smart" APM solutions. For more, Read APIs and CD: Rekindling Interest in APM - Part 2.