Observability is the ability to provide insights, automated analytics and actionable intelligence through the application of cross-domain data correlation, ML and AIOps across massive real-time and historical metrics, logs and trace data. Sascha Giese, Technical Evangelist Observability at SolarWinds, assesses the maturity of organisations when understanding the technology and what’s needed to progress this further, as well as the role AI can play in managing such processes.
Over the last two years, observability has gone from nascent tech to mainstream. How have your conversations with end customers about their understanding and need for observability evolved during this time?
Observability is basically just an evolution of what we used to call monitoring, but the requirements of the business have changed a lot. In the past, IT supported the business, but nowadays, technology runs the business. You can’t imagine a business without IT and with that, the requirements obviously changed. Modern solutions are required for modern technologies because there is more complexity in today’s modern IT environments.
In your opinion, is the CIO still required to be the primary champion of adopting observability or is this a technology now understood in the boardroom?
There’s a bigger problem which isn’t just limited to observability. CIOs fight for a bigger slice of the pie when it comes to budget and for most members of the boardroom, technology is difficult to understand. Often, taking a boardroom to a manufacturing plant and explaining the production process and product belt may seem like a more straightforward task. As a result, CIOs sometimes fail to show the value that their department brings to the business, however, they could usually use the exact same process and walk the other board members through the technology stack and explain what the individual elements do. It’s important to explain what might happen to the business if something fails – which would usually mean the business stopping for X amount of minutes. It’s therefore not just limited to observability, it’s a general problem for any CIO.
How can IT best build the business case for adopting observability platforms?
If you consider that we no longer use this three-tier application model where we had a database in the background, a front-end and what we used to call a dedicated desktop application or something that exists on a desktop. Nowadays, we look at decentralised applications which have a little bit here and there, so different clouds, completely hybrid.
We are now at a stage where it is very difficult for humans to understand how an application actually works or how the application delivery works right now. They need a solution to support them and make it relatively simple to find the root cause and points of failure. You can’t measure or fix these things without proper tooling and obviously, with the proper tool set. And it is very easy to create a business case around these numbers when you consider how much money you lose when you can’t produce or sell for 10 minutes.
How would you assess the maturity of organisations when understanding the technology and what’s needed to progress this further?
There’s no status quo in technology, we’re always moving and there’s always evolution. When we talk about AI, it’s more of a revolution.
When technology is moving on, it’s important for the business to keep up with the latest tech because if they don’t, their competitors will. Let’s say you own a shoe business and you respond a little slower than usual. As a customer you’d likely look elsewhere. Hence, it is very important to make sure that the technology stack you depend on works and that requires observability in the background.
For organisations that have legacy investments in traditional monitoring solutions, and that leverage native tools in cloud platforms, how can the migration to observability be made non-disruptive?
The first question is probably around whether migration is possible. If you look at sectors like finance or healthcare, you might find yourself in a situation where you use or depend on a solution written in COBOL and running on an AS400. Good luck migrating that to a cloud platform – that is next to impossible.
There are various reasons as to why these technical challenges exists and it’s important that an observability solution can deal with both; the old and the new, because you might not be able to migrate everything. In general, we have a lot of customers who are using our tools before, during and after a migration phase.
Technology is of course one part of the observability equation and equally critical are human resources. How is this balance shifting with the advent of a greater degree of automation and AI in observability and IT management?
AI frameworks in general are much needed across the whole tech stack. Today, we deal with so much data that we as humans can no longer understand it, and the time to process it takes a while. A machine, however, can do this a lot faster than us.
Deploying something like AI comes with challenges, one being concern around job security. Any invention in the past two centuries changed the way we work, but still left us with work to do.
AI is a tool like any other, but it allows us to have assistance and for us to outsource easy and repetitive tasks, which in turn gives us humans more time to breathe and more importantly, allows us to drive innovation for the business.
Looking ahead, what are your predictions for the future of observability and IT – how do you see SolarWinds evolving to meet these needs?
I don’t have a crystal ball, but we can guess that the amount of smart automation will increase in the future. We’re referring to AIOps, technology taking over simple tasks without human intervention.
At SolarWinds, we believe that AI should still be supervised. However, at some point, AI will have learnt so much and the more time goes on, the smarter it will be.
AI will also help the human decision-making process because it can access more data in a shorter time frame and offer suggestions which, as humans, we might need a week to comprehend.
It’s also possible to imagine something being introduced to the business where we can speak to a voice assistant and ask how much of our product we sold in Michigan in the last quarter, for example.
But, in order to make sure these things work, we need to keep up with observability.
