If you follow my blog, you probably know that I think that continuous delivery and the broader category application deployment automation are pretty great. I spend a lot of time working with engineers and architects who ‘get it’ but need some help selling the idea to their management. They need to build a business case, a return on investment (ROI). Researching this with some customers, I was able to put together a four page white-paper on release automation ROI. The basics:
Manual Deployments are Bad
Really, really bad. So bad that I’m recorded in this webinar [link] calling on them to die. Manual deployment processes are labor intensive enough not to scale. They are too painfully slow. Worst of all, they are naturally error prone.
From labor intensive to push button
Many teams I work with turn to automation when their deployment complexity gets out of hand. Whether it is because they have too many moving parts, too many servers, or just too many deployments a release engineering team can’t keep up at a pace that satisfies their customers (the business and development). To keep up, they would need to hire many more people just to run through deployment plans all day. Instead of incurring that cost, they invest in automated deployments where they can provide push-button deployments to other groups in lower environments, as well as production deployments for themselves. They can then turn toward higher value work of engineering zero downtime deployments and the like.
From slow to fast
Manual deployments are slow. This slowness causes inefficiencies in testing, and can impede our time to market. It’s pretty typical for an automated process to be 90-98% faster than a manual process. Long outages can be expensive. Shorter ones are cheaper and make more frequent deliver of new features more palatable.
From error prone to consistent
Manual deployments are error prone. Failed deployments to production can be disastrous. ROI calculations here often involve assigning a high dollar figure to a relatively rare occurrence and then estimate the savings by reducing the occurrence to something considerably more rare. In other organizations, deployment errors relatively frequently cause rollback windows and longer than necessary outages. Automation can resolve most of those problems.
From opaque to transparent
Manual release processes tend to be hard to track and audit. An automated process should also automatically generate detailed audit trails reducing audit costs and the risk of findings that need to be dealt with.
If you’re considering an application release automation project and are asked for a return on investment, you should read the full white-paper: The ROI of Deployment Automation