“There are two ways to look at analytics in an Operations environment:
• POST Cognitive
• PRE Cognitive
Post-cognitive is focusing your service management efforts on looking at events that occurred in a history after an outage. Post-cognitive is akin to driving down the road while looking out of the back window.”
Dougie Stevenson on Moogsoft.com.
There are two concepts relating to service assurance in Dougie’s article that I really like.
One is the pre-cognitive approach, which “takes in event streams in whatever form: partial, incomplete, even skewed across time, and produces in real-time clusters of alerts that are related (Situations).”
The other is “Situation Rooms, a collaborative UI where teams work together better and restore a service-affecting situation faster.”
The first is interesting because the use of machine learning is refining itself and producing insights based on streaming data rather than just post-event pattern analysis.
The second is interesting because cross-domain service impacts (situations) immediately imply collaboration across teams in most large service providers. The traditional way of handling cross-domain situations is via trouble-tickets and face-to-face collaboration. I haven’t seen Incident.Moog in action yet, but I do like the concept of assisted collaboration via situation rooms.
Taking it a step further, I also like the idea of using situation room simulations to generate training/learning rooms to get contextually relevant insights into your less experienced operators.
I’m not sure that I completely agree with Dougie’s analogy that using learnings from past situations is like driving while looking out the back window. Past situations do help to learn for future situations, but he’s right that analysis of past situations doesn’t necessarily help resolve the many previously unseen situations that will occur in the future.
Using machine-based learning to provide decision support to operators is clearly the target for next generation service assurance products.