There’s a lot of excitement around AIops, this can sometimes mean that it’s being deployed for the wrong reasons. Here are a few situations where AIops is contraindicated.
By David Linthicum
Artificial intelligence for IT operation platforms, better known as AIops, is an evolving and expanded use of technologies that for the past several years were categorized as IT operations analytics. The growth of AIops has been clear to anyone watching the market, but if you need some statistics, Gartner reports that by 2022, 40 percent of large enterprises will use AIops tools to support or replace monitoring and service desk tasks, up from 5 percent today.
That’s a pretty big jump. However, it’s also an indication that many enterprises may pick AIops tools for the wrong purposes—mistakes that will likely cost millions. Here is what I’m seeing.
Using AIops tools to fix bad cloud architectures and deployments. Those who haven’t planned a proper cloud solution for the business, and even an on-premises solutions paired with public clouds, are attempting to fix systemic issues. A poor plan will lead to performance problems and outages with AIops.
Just like “you can’t fix stupid,” poorly planned architectures need to be corrected before you apply AIops tools and use them properly. AIops tools work on the assumption that the solution’s configuration is sound before they can process alarms and resolutions properly. If not done in that order, you’ll just be teaching your AIops system how to correlate gigabytes of data coming from cloud and non-cloud systems, attempting fixes that are unlikely to be successful because they kick off other alerts and triggers.
Hoping AIops tools will eliminate people and costs. Those working cloudops now are, in essence, inventing a new discipline. Enterprises have seen the growth in cloudops specialists who are commanding some pretty good salaries. This has driven up costs, reducing the value they thought they would get by using public clouds.
I’ve seen enterprises invest in AIops tools based on a business case pointing to fewer ops team members needed, and the ability to automate costs out of the ops equation. Although there is a potential to reduce cost and staff much further down the road, AIops requires a great deal of ops expertise. You typically see AIops drive an expansion of the cloudops team, and costs initially rise for at least a few years. You have to invest in efficiency; you can’t remove dollars and expect good outcomes.
Using AIops to deliver better security. Those who do cloud security already understand that the ops automation processes is not a good way to defend your cloud-based applications and data. Indeed, the conflation of AIops with cloud security actually makes things less secure, considering that you’ll be dealing with security systems that are more complex.
AIops is one of those tools that needs to be implemented with great care in order to be effective. The market is moving at such speed that mistakes are going to happen; however, with a bit of common sense, you’ll find that AIops will eventually provide the value you’re looking for.
The full article was published here