If you think about many things you enjoy most in life – a good pair of shoes, jewelry, or a car, they all use different materials joined together in pleasing ways.
I'm looking at my Apple Watch right now, which has a beautiful juxtaposition of a rubber strap, metal housing, and glass face. It is the seams – how the materials are connected – which add to the beautiful user experience.
We see the seams as enhancing the user experience – can you imagine a watch made only of glass or rubber? In many cases, like intricate fabric or leather, the seams are accentuated to provide a more delightful aesthetic.
So why is it that in enterprise software, the moniker "seamless integration" has become prevalent?
With integration we should celebrate the seams that connect different technology stacks – in our case, SAP and the broader Enterprise Service Management context of ServiceNow. Calling this "seamless" would be a disservice to its purpose.
Why? Because to achieve true AIOps, those seams are critically important.
To understand this, let's first take a step back and look at the bigger picture. Our IT Operations customers tell us that their nightmare scenario is when the business informs them there is an issue. At first blush, you might think this is because it makes them look bad, but this couldn't be further from the truth.
Suppose a business user has to create an IT incident. In that case, it means they have been interrupted. The minimum cost of an interruption is 30 minutes – and that's assuming that the business users have something else they can do while they wait for service restoration. If the incident disrupts core business processes, the cost can be millions of dollars an hour.
Consider this: what is the ratio of incidents created by a machine and resolved before they cause business impact? We found that very few businesses even measure this ratio, and it is straightforward to get to 2:1 – i.e., over 66% of incidents created and resolved before a user notices. This is what we refer to as AIOps.
What is necessary to achieve this?
An accurate inventory of systems, services and data flows is the foundation of AIOps – in that it is impossible to identify issues if you don't know what you have. At Avantra, we find that most of our prospects do not have an accurate IT services inventory. How can you hope to provide an excellent user experience if you don't even have an inventory?
From there, it is necessary to define a baseline of what "normal" looks like via a set of domain-specific best practices, rules, forecast algorithms, and machine learning algorithms. Now anomalies can be easily identified and integrated into Enterprise Service Management and assigned to resolver groups.
Anomaly detection allows IT Operations teams to get far easier visibility into issues because most IT service incidents are compound problems that start with a simple problem and cascade.
A simple example: a batch job overruns, then a backup which should run after the batch job commences, and then we get to 9am, and the finance users start to post journal entries. The system slows to a halt, and service is impacted. Finance users start to raise tickets, and IT Operations begin to work on the model.
This example might sound banal, but examples like this are the reason why service interruptions occur. An anomaly detection system can quickly detect the expected runtime of the batch job and raise an IT service management incident if it overruns. An expert is already looking at the problem before the backup runs and causes further problems.
And this is exactly why the seams of integration are so important. None of this can happen unless you have a single view of Enterprise Service Management, like ServiceNow, and deep integration into complex enterprise software like SAP.
I say – let's celebrate the seams and focus on making enterprise software a pleasure to use.