In the 90s and 00s, when most enterprises decided to consolidate legacy custom applications into ERP, a Systems Integrator would turn up with a bus full of consultants, and they would get to work configuring it. You might be surprised to hear, however, that the ERP Operations team never left.
The problem of ERP data and operations silos lingers with them. Even in 2021, it's not unusual to have a team comprising hundreds of people whose sole purpose is to keep applications like SAP running, keep it up to date, and take care of security and audit needs. Because applications like SAP are notoriously difficult to integrate into IT Service Management and Operations Management tools, CIOs have left the ERP team as an island, separate from the rest of the Enterprise Software support team. It’s time for CIOs to get off of this island!
To start with, ERP is usually not integrated into centralized IT Asset Management, which is the foundation for an integrated operations strategy. In the case of SAP, customers use a separate closed-wall asset solution called Landscape Management Database (LMDB) which is maintained by the ERP team and contains only SAP systems, with no capability of automatic discovery.
ERP teams then often have their own IT Operations Management tools for incident management, and they take service desk calls from the business that log when the ERP software is not running as expected. That's right — many ERP incidents get first reported by a business user who cannot perform their function.
Take the real-life example of a senior executive who has two parallel situations. First, the organization has thousands of scripts which keep systems running day-to-day, with limited change management or compliance ensuring they keep the business safe. And second, she has described how she typically finds out about a serious incident from the CFO. No IT executive wants to be blindsided by the CFO - it’s indicative of a broader malaise.
Can AIOps for ERP help?
Many workflow vendors hype AIOps as the solution to operational logjams. But does this apply in an ERP context? And is AIOps the term we should be using here? How does this differ from automation efforts of the past?
The first thing that AIOps can offer is discovery, or "Observe." We find it is typical that customers don't know how many ERP systems they have, their function, and whether consultants followed best practices during the initial configuration. Machine learning algorithms are applied to understand anomalies within the environment.
From there, all of this information flows into centralized IT Asset and Service Management tools like ServiceNow, known as "Engage." AIOps automatically identifies anomalies (e.g., a connection to a bank is not functioning correctly) and creates correctly prioritized incidents assigned to the resolved group best suited.
Since the priority is known, escalation paths are automated, and the relevant senior engineers and managers receive notifications.
The real value in AIOps for ERP is that historical incidents and knowledge bases are a treasure trove of automation possibilities. In many cases, the issue causing a fault is well known; for example, a security vulnerability released requires a configuration change. In this case, you can identify a check (Observe), communicate this to a business user (Engage), and then move onto the third pillar of AIOps — automatically setting the parameter (Act).
The impact of change management
All of this is very well, but organizations define maintenance windows for this type of work. The frequency and duration of these windows depend on the function of a specific system. For example, some critical systems have a 30-minute maintenance window once a month.
As a result, organizations orchestrate change in an orderly manner, and this is where the potential of AIOps comes into play. The changes in a specific change window have been tested and approved. An AIOPs platform can shut down the relevant systems in an orderly fashion, apply the changes (in parallel) and bring the systems back up again.
The change window might be for 30 minutes, but the platform can perform complex tasks in a matter of a few minutes, returning a business-critical system to its users ahead of time. What's more, the root cause for most problems caused in maintenance windows is operator error. AIOps systems are not infallible, but they perform the tasks requested correctly.
The reason ERP is so complicated and poorly understood is that businesses have been content to leave those systems on their island. But you can forget about workflow automation until this problem is reckoned with.
It may now be time to rethink this, because the tools and platforms now exist to achieve a semi-autonomous IT operations environment and integrate applications like SAP into your centralized IT Operations processes.
For more on how this plays out in SAP environments specifically, check out research paper, How Automation Accelerates Innovation for SAP Operations Teams.