Since getting such great feedback on last year's APMdigest 2017 prediction analysis, I wanted to highlight some of the predictions made at the start of 2018, and how those have panned out, or not actually occurred. I will review some of the predictions and trends from APMdigest's 2018 APM Predictions. I will group together concepts as Pete Goldin (editor and publisher of APMdigest) already does this with the predictions. I enjoy reading these and thinking about what exciting things will happen over the upcoming year. I wish everyone a happy new year, and hope we all have time to reflect and think about our exciting industry.
There is no question about the bright future of the public cloud, and how most enterprises are further reducing data center footprints. Similarly, new paradigms to manage applications are coming in the form of containers along with Kubernetes being the next application “operating system”. These trends do not mean there has been a significant reduction of the usage of legacy systems in the enterprises today. In fact, many of these new applications further increase traffic to legacy systems, most often via APIs.
Adding these layers causes a lot of challenges in operations as each technology comes with tools necessary to manage and diagnose performance problems. This complexity is increasing very quickly, which has created the need by buyers for tools to help reduce the manual work both in terms of automation, root cause analysis, and pattern recognition. There has been a lot of hand-waving (marketing … “look over here”) regarding advancements in AI and automation, but were the 2018 predictions accurate?
At least 15 predictions spoke of improvements in AI and AIOps. And while we certainly saw a lot of marketing and press releases about AI from APM players throughout 2018, there were some new capabilities and certainly a vision. There was neither major advancement or any kind of tie-in to automation or remediation, very few are generally available, and most are rules based engines which is a far cry from a real AI. While automation and closed-loop use cases are clearly the direction Gartner and others predict regarding AIOps, we are a long ways from that point with very select use cases being possible today.
The use of AI can also be negative as there is a lack of context. This wave continues to grow, and in 2018 we started to see this prediction come true, I’d rather let the words of Mehdi Daoudi, founder and CEO of Catchpoint, do the talking with this great prediction:
In 2018, we expect to see a growing realization of the limitations of today's AI for IT issue identification and resolution. As the number of performance-impacting elements (and IT complexity) increases, AI can be helpful in identifying some problem spots, but human intervention will always be needed to discern what (if any) issues are truly customer-impacting and thus warrant a call to IT teams in the middle of the night. For example, let's say a front-end server is slowing down. Are customers growing angered? Are revenues in danger? Or can the issue wait until the morning? These are things that a machine can't necessarily learn. AI without guided human intervention can actually have the adverse impact of desensitizing IT staffs and making them less effective.
APM Market Shifts
APM tools still represent the best data source for IT to get business visibility, and for many CIOs it’s the most important data set to understand how their teams are delivering business outcomes. There were many predictions along these lines and generally this has increased as the APM market size continues to grow.
Customers would like a single pane of glass, unified monitoring, or other convergence, but the reality of the fragmented monitoring tools still exist. Multiple tools are used, and that’s unlikely to change anytime soon as the demand and innovation for new technology outstrips the need to consolidate tools, which is generally a cost-saving exercise with less upside. There are many vendors trying to state this is happening, but it’s not as fast as new tools are entering the enterprise to meet ever-expanding technology stacks. Those who predicted convergence were not at all accurate, in fact, we have more tools than ever before, especially with open source becoming more relevant.
The APM ecosystem did make significant advancements this year (that one was my prediction). Vendors introduced new product offerings beyond their time series metrics and log analytics tools. They all have incorporated tracing, and of course, there is more to APM than tracing, but it’s one of the fundamentals. These offerings all leveraged and incorporated open source tracing based on OpenCensus and OpenTracing, most of which are also using aspects of Zipkin’s protocols and designs as it’s the most mature of the OSS tracing tools. This allows these vendors to have the beginning of an APM product without having to build the agentry or depth of many other leading tools. I do not believe these solutions are gaining any significant adoption yet, as the lift is too heavy for most enterprises. The maturity has a long way to go, but it will get there in the next several years.
Unfortunately, the market at the end of 2018 looks the same as it did in 2017 for end-user experience or digital experience monitoring technologies. There were a few innovations in network and internet monitoring along with new APIs going into browsers for monitoring, Chrome Dev Tools made huge advances this year. If you haven’t looked at it recently, just fire it up. Thankfully, there has been increasing adoption of these technologies by end users, as the benefits continue to be very clear.
More than half (61%) of respondents reported that their teams are practicing observability, an 8% increase from 2020, signaling that overall adoption is on the rise, according to a 2021 survey from Honeycomb. However, the majority of respondents indicated their teams are at the earliest stages of observability maturity ...
Respondents to an OpsRamp survey are moving forward with digital transformation, but many are re-evaluating the number and type of tools they're using. There are three main takeaways from the survey ...
More and more mainframe decision makers are becoming aware that the traditional way of handling mainframe operations will soon fall by the wayside. The ever-growing demand for newer, faster digital services has placed increased pressure on data centers to keep up as new applications come online, the volume of data handled continually increases, and workloads become increasingly unpredictable. In a recent Forrester Consulting AIOps survey, commissioned by BMC, the majority of respondents cited that they spend too much time reacting to incidents and not enough time finding ways to prevent them ...