APM, Analytics, and Industry Trends
Just read Gartner Q&A: Analytics vs. APM with Will Cappelli, Gartner Research VP in Enterprise Management, part one, part two, and part three about his latest report: Will IT Operations Analytics Platforms Replace APM Suites?
The discussion is very interesting and informative, touches areas that interested me for a long time, but the title doesn’t make sense for me. Maybe I just got behind in terminology – but here is as I understand it.
I joined Hyperion in 1999 and, with one break, did, in a way, performance management of performance management applications. I was responsible for application performance of business performance management and business intelligence applications. At this point you probably can guess some questions that came to my mind. Can we use anything from application performance management in business performance management? Or, vise versa, can we use business analytic software for application performance analysis? Well, I haven’t still figured out a meaningful way to do it.
I am also involved in CMG (worldwide organization of IT professionals specializing in performance, capacity and service management), where there was an idea that skills of application performance and capacity management may be used for business performance and capacity. I waited with interest if the best experts in application performance and capacity management (and CMG is the place where you find them) would be able to come with interesting ideas of using their skills for business (beyond the trivial fact that performance impacts business). Still waiting.
While these two areas, application performance and business performance, look very similar and, of course, often use the same underlying approaches and math, the devil is in details. They are just two different areas and not much may be re-used between them (as soon as we get to details).
Reading the discussion, I came to one explanation of what is going on. It also explains one more phenomenon surprising me for a while. The phenomenon was that there are many new companies on the market providing only end-user monitoring (EUM, or real-user monitoring, RUM) and apparently doing pretty well. While EUM is definitely a very important APM area, my understanding was that it is just one part of APM and, without other parts, provides very limited APM value. My understanding was that we should rather move toward integrated APM solutions. Still the impression is that some such EUM companies doing even better than companies providing integrated APM solutions.
Well, the explanation is that web analytics (even with EUM) is just a completely different area from APM (although EUM is an important part of APM too). Web analytics is a business application (and yes, end-user performance is one of components of business analysis now) and APM is an “IT” application. If we agree on this, it explains all these strange titles about death of APM or replacing APM by analytics. End-user performance with web analytics became a business intelligence application – and the business intelligence market is many times larger than IT intelligence market (where we still have all true APM solutions – with all their problems because the APM is, in essence, much more sophisticated area technically than business analytics). It was always this way – recall market size of business intelligence (Hyperion, Cognos, Business Objects, etc.) vs. market size of companies providing monitoring and deep diagnostics solutions.
Business intelligence/analytics moved to a new level: from sales analysis we are getting to user clicks analysis. And there we have another level of data volume (==big data) and we are getting to the point when end-user performance is just one more dimension of business data. Performance impacts business – and business wants to analyze it. Well, in application performance management we worked with “big data” for ages – and have you heard about it as a world problem until it got into the business realm?
But nothing changed in APM – if we find a way to separate these business-related applications, I guess we find that the APM market is slowly growing and that APM applications are slowly becoming better and better. To manage your applications, you need an APM solution, not web analytics and not EUM only (although EUM is important for APM too).
As I highlighted in my performance requirements article (here is the last version just presented at CMG’12), we have two completely different views: business view and “IT” view. Business is interested in one (and only one) performance metric – end-user response time. That’s it, nothing else at all. All other stuff – throughput, resource utilization, bandwidth, latency, etc. – is for IT internal use only (even if “IT” is the core of the company as with many web services). Yes, business people know these words too – they hear them all the time as explanation why we can’t get requested performance or why we need to spend another pile of money on something IT needs – but if “IT” delivered the requested performance they were quite happy not even knowing these words. And that is exactly what we see here: we have solutions for business (EUM – which is much simpler of integrated APM solutions and provides everything business needs) and solutions for “IT” (with all these stuff you needed to manage applications performance – but much more sophisticated and difficult to implement).