Introducing Machine Learning into DevOps
Having been intimately involved with running Managed Services teams & platforms over the years I don’t think any of us are really huge fans of the “get out of bed” calls at 3am.
Without going into the what and why they happen in any great detail, the common sins and problems that do occur are usually the simple things that are well understood. My favourite sin of them all is - disk space running out!
This is the one that likes to happen in the small hours of the morning when we are least expecting the alarms to go off, it’s a great simple example supporting why we have created Intelligent Operations aka IntelligentOps.
So what’s the answer?
Once you have taken your managed Cloud environment as far as it can go with traditional automation, and when as many of the laborious, repetitive manual processes and human decision making points have been absorbed into your platform, the natural and next evolutionary step is now playing out beyond the age of DevOps.
This is the introduction of Machine Learning (ML) into DevOps; remove the human element further - no more operational blindness to repetitive alarms and missed alerts. Machines with the right data can and will take action and make the right decisions. We are working on fundamental self-healing and self-tuning with the next order of differentiation being an extension into the realms of cost optimisation, adaptive architecture and business intelligence.
We don’t worry about disks running out at 3am anymore, or those repetitive maintenance problems that keep cropping up even in well automated, mature systems. IntelligentOps will notify all concerned, gracefully put the platform into a fail-safe and user friendly maintenance mode, resize the disk (or perform any other well understood DevSecFinOps activity) and when operations are complete, bring the platform back online. Whilst this may be a rudimentary operation, it perfectly articulates the benefits of incorporating ML into our next generation managed services platform.
In turn, engineers that don't have to get involved in well understood operations translates to savings to your operational budget and an improved operational focus on the areas within your business that add value.
For more information about our Cloud Managed Services and free on-boarding to AWS click here.
In my next blog post I will provide more insight into how IntelligentOps works under the covers, the use of predicative analysis and anomaly detection, the concept of bots and the extensible framework we’re using to deliver industry vertical solutions...