The definition of “sprawl” is to spread or develop irregularly or without restraint, and can have either positive or negative connotations. However, in the IT World, ‘Tool Sprawl’ is considered something to avoid. . It’s why “fixing tool sprawl” always seems to appear on the list of things to do. At best, it’s difficult to maintain – at worst, it can lead to overlapping solutions and overspending, all while creating additional silos throughout IT.
If you’ve worked in IT at all, you’ve discussed tool sprawl with somebody, usually in the context of “stopping it” or “fixing it.” Sometimes the topic is labeled “tool consolidation” which often results in buying rules such as emphasizing “single vendor” solutions over best-of-breed point solutions. While those solutions have their value, they should not restrict a team’s ability to find the solution that best solves their problem.
Instead, AIOps is the remedy to fix some of the negative effects of your tool sprawl.
A few caveats:
- I’m not implying that AIOps actually STOPS tool sprawl. Instead, let’s discuss how AIOps can help overcome the silo problems that result from tool sprawl. .
- If your AIOps solution is missing a key technology, then your tool sprawl will worsen
Why Tool Sprawl Occurs
If we’re going to talk about how to “fix” tool sprawl, it’s important to look at how and why tool sprawl occurs in the first place. The shortest, simplest answer is simply “innovation.” New technologies emerge that make it better / easier / quicker / cheaper to deploy business systems and/or applications. As these new technology stacks emerge, tooling around them also evolves – anything from operational planning to deployment to monitoring.
Thus we come to the IT conundrum we discussed earlier (buying new point tools or waiting for your “suite” (quotes intentional) to add support for the newest technology – and HOPING that whatever gets added can be somewhat effective.
The end result is practically guaranteed. You will have a collection of disparate systems throughout operations that must be maintained, monitored and analyzed for improvements.
Making Sense of Your Data
One of the bigger issues with tool sprawl is the Petabytes of disparate data sitting around from all your tools. There are so many differences – Different data types; different data storage rules; different meaning; different timing. This leads to the need to have sets (dozens? hundreds?) of trained power users for each system to interpret the data and make the appropriate decisions.
Ironically, many of these tools promise to break through silos within their specific system, but end up creating their own operational silo.
Unsupervised Machine Learning
Spend more than 5 minutes looking at IT solutions (or any technology for that matter), and you’re sure to come across product descriptions containing AI (Artificial Intelligence) and ML(Machine Learning). As I’ve discussed before, not all AI and ML is created (or performs) equally. One big difference is whether or not machine learning is unsupervised.
Here’s the important take-away from the video – when you have unsupervised ML, you can add valuable analysis to practically any data stream. More importantly, it also means you can effectively analyze data from different datasets together – uncovering unknown relationships between systems and gleaning information that can help you become more predictably proactive – eliminating a number of problems altogether.
The Light IN the Tunnel
There is hope – there is a light, but let’s be clear. It’s not eliminating tool sprawl. Instead, put your efforts into making better use of the datasets you have. After all, you’ve spent a great deal of time and effort evaluating and deploying your tooling. Instead of simply throwing away your judgment and/or the value you get, leverage that data.
And this is where AIOps CAN help. You can probably guess what I’m about to say next:
If your AIOps tooling is based on unsupervised machine learning, then your sprawling datasets can become an operational power house. Conversely, if your AIOps tool doesn’t utilize unsupervised machine learning at its core, then it will simply amplify the problem you already have – trying to find which pieces of your mountains of data go together, and then trying to understand what it means to be related. One of these tools will simply toss another tool onto the pile of disparate tooling.
So STOP TRYING TO STOP TOOL SPRAWL. Instead, enable your teams to properly analyze all your data together and embrace the power of expert data.
