Don’t Let AIOps Hyperbole Detract From Hybrid IT Success

The following post first appeared on the Intellyx Brain Blog.

The world of IT operations has changed dramatically in the last decade. In simpler times, say, way back in 2005 before the cloud, tracking down issues in production environments was relatively straightforward.

If some problem cropped up, the first question an ops troubleshooter would ask would be ‘what changed?’. Was there a recent deployment? Did some server just run out of memory? Did the janitor unplug the router again? The conventional wisdom was that one effect had one cause, so find and fix the cause, and voila! problem solved.

For all of you ops veterans reading this, I’ll admit I’m oversimplifying things – but regardless, there’s no arguing with the fact that keeping today’s production environments running at top performance is nothing like how people did it a decade ago.

The difference, of course, is that today, there are too many variables – too many possible causes across multiple environments with complex interdependencies. And on top of all that, everything is always changing.

Welcome to the world of hybrid IT.

Is AIOps the Hybrid IT Panacea?

Hybrid IT refers to some combination of one or more public clouds, private clouds, on-premises virtualized environments, and on-premises legacy environments – keeping in mind that every organization is different and thus has a different combination of one or more of these environments.

But hybrid IT is more than just a multiplicity of environments. In reality, it’s a workload-centric management approach that abstracts the underlying environments, giving organizations the ability to make deployment decisions based upon the particular needs of the workload – especially when those workloads serve customers.

Earlier generation IT operations management (ITOM) tools clearly can’t keep up with this new hybrid IT reality. Enter AIOps – a new category of ops tooling that applies artificial intelligence to the massive streams of data that the systems, applications, and networks generate, in order to better identify root causes of issues and ideally, predict them before they occur.

The basic AIOps story goes something like this: take these various ops data feeds (which ops people snarkily refer to as data exhaust), feed them into your AIOps tool of choice, and out will come either (a) insight into the root causes of problems, or if you’re lucky, (b) some kind of useful prediction, say a heads-up about a problem that is brewing but hasn’t actually caused a serious issue yet.

Both outputs can be useful, to be sure – but they are essentially point solutions to point problems, and IT Ops is rarely if ever about dealing with point problems, especially in the modern hybrid IT world we live in today.

AIOps Built for Hybrid IT

However, not all AIOps products are the same. If hybrid IT were simply a mélange of environments, then AIOps tools that provided greater insight into running apps or systems in one environment or another might be sufficient to give ops people what they needed to satisfactorily manage the entire ops landscape.

But it’s not. Because of hybrid IT’s workload-centricity, any AIOps tool worth its salt has to take a holistic approach, pulling together the appropriate insights from across different applications, systems, and environments to provide useful insights for ops teams as they deal with the knotty interdependencies that hybrid IT brings to the operations landscape.

Furthermore, such holistic insights must necessarily take place in real-time – both because problems come and go with appalling frequency, especially as companies move to containerized, microservices-based applications, but also because the underlying technology itself is always in flux.

Trying to gain insights after the fact from old data sets is worse than useless because it sucks up the resources of the organization. But gaining insights into a single aspect of a complex enterprise infrastructure to the exclusion of the big picture can also steer ops teams down a rat hole.

OpsRamp: Beyond ‘What Changed’

Any AIOps tool that fundamentally tries to answer the question ‘what changed?’, or even ‘given what changed in the past, what’s likely to happen next?’ is simply working on too one-dimensional a level to address the full spectrum of modern IT ops challenges.  Not all AIOps vendors are cut from this one-dimensional mold, of course. One vendor who takes a multi-dimensional perspective on AIOps is OpsRamp.

Fundamentally, OpsRamp takes a service-oriented view across network, application and hypervisor topologies. It provides better access to data feeds than other AIOps tools, including data from all existing point tools as well as infrastructure elements and endpoints on the edge. OpsRamp discerns insights from the data holistically, rather than each feed separately.

This holistic control also enables OpsRamp to provide real-time insights that help ops teams contextualize issues – where the context is the particular hybrid IT reality that is salient at a particular time for a particular problem.

And because OpsRamp works its magic across environments, it can bring interdependent organizational silos together, with the result being better business outcomes for the organization and its customers.

The Intellyx Take

Of course, not all ops problems are so complex and intractable – although those tend to be the problems that get the bulk of the attention. In reality, most problems are simple and have simple solutions. For such problems, OpsRamp can fully automate the solution, cutting the human entirely out of the loop. No longer any need for L1 support.

The automation that OpsRamp provides frees up ops personnel to focus on the more complicated problems – problems that require sophisticated teamwork across silos, typically by bringing application development and ops together (‘DevOps’), or adding business users to the mix (‘BizDevOps’) in order to collaborate to resolve issues.

Many enterprises are already well down the DevOps or BizDevOps roads already – so it behooves them to select an AIOps tool that supports both automation and collaboration in equal measure, breaking down the organizational silos that impede rapid resolution of difficult problems as organizations fully embrace their hybrid IT reality.

Next Steps:

Service-Centric AIOps White Paper


Recommended posts