29 July 2020
Reasons not to automate your IT service delivery
by Mark J Menger
We shouldn’t automate everything
- We do it infrequently
How important is it to perform the activity rapidly, consistently, and cost-effectively when the time comes to do it? How much do you want to pay in time and labor to rediscover/recreate the knowledge necessary to perform the activities?
- It’s costly to automate it
It is likely helpful to perform a net present value (NPV) analysis of the knowledge creation investment. You may be surprised by the result.
- It’s cheaper to send it overseas
Who should own the knowledge of how to operate your business? Suppose NPV and risk management analysis suggest that handing off accountabilities to a third party, possibly overseas is a wise business decision. In that case, you may want to set the expectation that the third party creates version-controlled automation assets for the activities you’ve outsourced to them. These assets should create financial efficiencies for the third party and reduce the cost of migration if you choose to in-source the activities or move them to another third party.
- I don’t have the time
We all have limited capacity to deliver the results for which we’re accountable. A question to ask is where you want to spend your and your team’s time. Automatable activities are often low-value labor-intensive (consider this when performing NPV analysis). Even more concerning is that these low-value activities often create risk for high-value activities when delivered inconsistently.
- We don’t have the budget
If your business has ever-increasing needs for rapid time-to-value in concert with improved reliability, then you likely need to create near-term financial capacity for speed and consistency improvements. And, to maintain fiscal responsibility along with your automation efforts, align them to business-value-facing metrics. If you’re not helping move the dial measuring a strategic business metric, then re-focus the effort.
- It’s not possible
Super creative folks are continuously churning out solutions to previously intractable problems. There are IaC and SD* solutions that didn’t exist two years ago.
- We tried it before
see above
tags: big-ip - automation - hashicorp - terraform - devops