The Missing Step in Your AIOps

Jeff Drury
12/11/2023
min read

Watch out “Cloud,” your unholy reign as the most misunderstood I.T. buzzword could soon come to an end.   

I.T. management loves a good buzzword that simultaneously inspires, confuses, and drives staff crazy.  With “Cloud” getting a little old and unfashionable, “Observability” and “AIOps” are ready to take over.   

Early in my I.T. career as a help desk tech, I joked that I was responsible for support of anything that plugged into an electrical outlet.  The promise and buzz around observability is close to that broad and unmanageable scope.  With the concept and promise of observability, I.T. can have an all-seeing, all-knowing vision of anything that happens in a business and bring that back to the mythical single pane of glass to guide business decisions in real time.  AIOps takes the concept of observability further and promises low-to-no effort recognition and remediation of our worst finger pointing, war room issues before us lowly humans can even fire up a bridge.   

 In reality the concept of observability is amazing, but the implementation is held back by the usual suspects:  Organizational silos and manufacturers refusing to cooperate stand in the way of achieving true observability.  This is not a new problem in the technology world.  We only need to flash back to the 1990s and the rise of the corporate network.  On the physical front we had crazy connectivity options like coax, twisted pair, and eventually ethernet.  Adding to the complexity, I.T. titans like Apple, Novell, and Microsoft backed their own communication protocols like Apple Talk, IPX, and TCP/IP.  A plucky west coast startup named Cisco created a magical little box and backed industry standards for communication.  That little box, called a router, could connect to all the goofy physical cables, often through an equally goofy 25-pin AUI module, and then efficiently move all of those 1s and 0s between the protocols, and even to other networks… BOOM welcome to the Internet.

Fast forward to the problem of Observability - every device, application, or web enabled toothbrush creates telemetry, Metrics, Events, Logs, and Traces, in their own proprietary and most of the time arbitrary format.  Sound familiar?  The more problems change, the more they stay the same.  Cisco is back with a different kind of box, this time a platform, that brings in all of the telemetry and creates meaningful relationships between the data, in real time.  This time the industry standard is Open Telemetry or OTel, a uniform method for creating and classifying telemetry.   

Once all the data is formatted to a standard and linked together to form meaningful correlations, AI can step in to see patterns and respond accordingly.  Everybody wants to go from four 9s to five 9s of availability in their business.  With traditional people, process, and data making the jump to five 9s is extremely difficult.  With true observability and a little help from AI, the process is a whole lot easier.  I for one welcome our new AI overlords.

Jeff Drury

Related posts

Keep exploring!

Why Data Science is Critical to Your Business

Revenue-driving, cost-saving decisions require decisions based on intelligence, not assumptions.

Dr. Jim Sullivan
10/15/2023
7
min read

Building on Sand or Stone? Four Essential Lessons for Establishing Data Strategy as Your Enterprise's True Foundation

Four real-world lessons about your data strategy.

Asif Haider
9/30/2023
7
min read

Are We Safe From the Dangers of AI?

While movies may scare us with out-of-control robots, our experience with AI tells us a different story. There are many benefits to be found in safely embracing AI.

Carrie Fields Amy
10/20/2023
7
min read