The Muddy Messaging of Observability and Application Performance Management
Here’s a question I get asked quite a bit: “How is Enterprise Observability different from APM and/or just plain Observability?” It’s a reasonable question, given the muddy messaging that exists in the market. When I began my Instana journey, our solution was called “Dynamic APM.” Whenever we spoke to potential customers, we acknowledged that we were delivering much more than just “APM,” but we weren’t sure what to call it yet.
Fast forward 4 years — now we know — ENTERPRISE OBSERVABILITY!!!
When you hear it, you nod your head and think, “Ahh yes, but of course!!!”
But just because we finally figured out what to call it doesn’t mean that it’s obvious to everyone else just what Enterprise Observability means, nor how it’s different from just Observability, or APM (Application Performance Monitoring) for that matter. Let’s take care of that, shall we? First let’s look at what Enterprise Observability is NOT.
Observability is NOT APM v3
Along the way, we thought about calling our solution APM Gen 3 or APM 3.0. While capturing the concept of helping organizations manage their application on the next generation technology stack, it was not completely accurate for several reasons. The one that sticks out most to me, though, is the number of users.
As we spoke with more users (many of them owning a Gen2 APM tool), it became clear that dealing with the Cloud Native Technology Stack required a different approach. Even more glaring, though, was the exponential growth of stakeholders in production application performance – and how the legacy APM tools were struggling to serve this new group of application monitoring users.
My favorite example was the online retailer who had invested a LOT of money in their APM 2.0 tool. The tool did what it was supposed to, but as the organization moved to agile development and began accelerating update frequency in a CI/CD pipeline, the performance management tool just couldn’t keep up with the pace of the Dev team. While that particular scenario wasn’t unusual for us, it was the user count that caught my attention. About 10 IT Ops team members used the Gen2 APM tool, each user going through weeks (or months) of formal training and years of on-the-job experience. Every time the tool was needed to see an issue and/or troubleshoot it, these 10 individuals were the only ones who knew how to interpret the data they were seeing. During the two-week trial of Instana, more than 250 individual users logged in. When the company became an Instana customer, the total user count reached 1,000 with hundreds logging in every week.
This scenario has repeated itself over and over at customers who have moved from a legacy tool to Instana: 400, 800, 1200 unique users logging into Instana – all able to use the tool with no rigorous training program or specialized expertise on how to interpret the data they’re seeing.
Metrics, Traces, and Logs
Even as the exact definition of observability remained flighty, a general acceptance came about of “something” built around M/T/L
- M – Metrics (or Monitoring)
- T – Traces
- L – Logs
Some like to include a fourth bullet – Events – but in reality, Events are [mostly] captured through Logs anyway. I’m not sure those that try to claim events separately aren’t just trying to get a vowel into an acronym.
Observability Is More Than Pretty Charts
As we were preparing to launch Enterprise Observability, a prominent editor asked me what I thought about the new “observability attitude.” He elaborated that there was a school of thought that you shouldn’t deploy ANY monitoring. Instead — just gather your log data and create custom dashboards with Grafana (side note: Instana has a plug-in for Grafana).
I gave some serious thought to his question — because I was a little shocked at the premise. Ultimately, I answered this way — the point of application monitoring tools in general is NOT to just expose response times and loads on pretty charts. I mean, if all we wanted to do was throw some log metrics onto a dashboard and call it a performance solution, we could do that in no time. As we will see, monitoring is PART OF ENTERPRISE OBSERVABILITY, not something separate.
The point of these solutions is to SOLVE PROBLEMS (when they occur). And when they occur is the sticking point – they can occur at any time!
Andddddddddd, they can be caused by almost anything!
The only way to ensure you have all the data you need to SOLVE PROBLEMS is to select the appropriate application performance monitoring / management tool.
Differentiating Monitoring, Observability, and Enterprise Observability
Ultimately, the idea of not deploying any monitoring into a critical application is a non-starter, especially as teams accelerate their deployment pipelines. If you look at ALMOST any definition of Observability, you see the M stands for monitoring. Thus, as teased earlier, Observability begins with some level of monitoring.
Add in a dash of individual trace collection – and throw in a sprinkle of log information (from lots of sources), and you get the primary definition we talked about before.
Which brings us to Enterprise Observability. What makes Enterprise Observability so different?
Well, I think it’s important to point out that Enterprise Observability HAS Metrics, Traces and Logs. Further, the metrics are the deep metrics you expect and need from Application Performance Monitoring. So out of the gate, Enterprise Observability has the breadth you want from Observability solutions and the depth you need from APM tools.
That IS a powerful combination, but it’s not JUST the data that makes Enterprise Observability different. It’s how that data is retrieved and used that makes a difference.
When Instana launched the first version of our fully automated APM solution, one reporter commented that everybody knew you should automate monitoring. Yep, everybody KNOWS it’s needed – but not everybody DOES (or CAN) automate the monitoring lifecycle. When we say automation, we mean every – step – of – the – process:
- Change Detection
If you aren’t automated throughout the entire monitoring lifecycle, then you WILL slow down your pipeline every time you roll around to the monitoring piece.
One of the more powerful features of Instana is the Dynamic Graph. This comprehensive application / infrastructure model is the jet fuel to our Observability Engine. It allows an understanding (both programmatically and to human users) about how every service relates to every other service, infrastructure, end users, web sites, data stores, etc.
It drives our ability to tie independent applications and/or services together because they share a common upstream request or a common downstream asset. And we do it all automatically, which feeds the third leg of Enterprise Observability – intelligence.
From a marketing perspective, it’s not simply intelligence – it’s taking intelligent actions based on the data that’s ingested and exposed to the end user. It’s the combination of always-on automation and the ability to recognize context of every piece of data, coupled with a machine-learning driven AI and analytics engine, that not only allows Instana to perform automatic root cause analysis and identify triggering events or problems, but also provides the right data and context so that application stakeholders can make better decisions themselves on how to react.
EASE OF USE
While there can only be three legs to a stool, there’s an underlying concept of how easy it is for ANYBODY to get to the answer. Anybody from Dev to the Business can perform root cause and routing for proper handling – without days (or weeks) of intensive training. Each stakeholder gets the data they need visualized in a way that makes sense to them, so that there’s broad use of the tool – viral, almost.
That’s what makes Enterprise Observability the most enterprise-y of them all – the fact that in days, hundreds of users can log in – even during a trial – and understand what they’re seeing. And that’s powerful enough to make any application owner happy.
Skeptical? That’s cool – if you want to see for yourself how quickly Instana can add value to your organization, try it for yourself with our full FREE two week trial (remember, no limit on users). Or if you just want to see how easy it is to use overall, try out our Observability / APM Sandbox.