Thoughts about APM Predictions
One prediction in 13 APM Predictions for 2013, surprised me: 7B. Splitting the Market – Ops vs. Dev Tools by Chris Neal, BlueStripe CEO and Co-Founder (@ctneal).
My understanding of the APM market is that it is not yet a mature market. There are a lot of quite different products on the market whose vendors name them APM or whatever nice buzzword is around (some changing them quickly according to the latest fashion). Some are ‘deep diagnostic’ tools which provide insights into specific environment, some are transactional tracking tools, some are rather traditional monitoring tools, and some are Web end-user monitoring. When I started my collection of links 7 years ago, I classified them into Application Management Tools, Enterprise Monitoring Tools, and Transaction Tracking/Business Activity Monitoring – definitely far from ideal (especially because APM became an umbrella category), but still looking for a better way to do this. But marketing messages of almost all vendors were practically identical: “Buy our solution and you get full insight into your environment with minimal configuration and overheads”. Which was quite far from the truth on all three points.
It became much better (so I am surprised by the posts talking about APM death basically because APM doesn’t deliver on above mentioned promises – well, it was much worse some time ago). And now, I guess, we see some symptoms of maturing. Gartner’s five dimensions of APM did, I guess, a good job of structuring the solutions (at least in what concerns ‘Monitoring’ – it doesn’t look like we get to ‘Management’ yet). So some vendors started to deliver a more meaningful message on what their solution really does instead of saying that it solves all your performance problems. And, in a way, some tools started to be grouped around these two categories – dev-oriented vs. ops-oriented. See, for example, The Third Generation of APM by Bernd Harzog (@Bernd_Harzog) – although Bernd is using rather un-orthodox terminology there.
So I see a point in Chris’ prediction if we talk about maturing market and more concise vendor messages about what their solutions actually do (and maybe grouping them around these two categories). But if we talk about overall long-term trend, we probably expect APM tools to cover all this functionality in one seamless solution (with all five Gartner’s dimensions and, probably, much more). That is the main problem of the APM market – even the companies that formally cover all this APM functionality have actually several separate solutions not well integrated between themselves. And I suspect that the first vendor that comes out with really good integrated solution will get a great competitive advantage. Almost all other APM predictions actually screaming about this market need – having one, comprehensive, end-to-end solution (either by one vendor or by cooperation of several). And DevOps need that from APM: not only ability to register a long transaction, not only ability to see at what component the time was spent, but also tracing it down to specific query or method and its parameters – because otherwise you are actually back to the traditional approach when you try to reproduce operational problems in test or dev environments and thus lose a significant part of the promised APM advantage.