While not necessarily environmentally sound, our consumer environment these days sometimes makes replacement a sounder financial choice than repair. A summer grill or a microwave oven is so relatively inexpensive that anything beyond surface repair is usually cost-prohibitive. If it doesn’t work, pawn it off on your brother, take it to the summer cabin, or simply to the dump.
Within IT, we’re often afforded the same luxury without the environmental cost. In fact, rebuild over repair should not be an accident, but a deliberate strategy with IT services. All too often have we seen firsthand how IT-organizations when facing a serious incident start to try to roll forward out of the trouble. This in turn leads to unnecessary delay and we often paint ourselves into a technical corner that doesn’t allow rollback. Perhaps better would be simply to skip the fix and rebuild.
Applications and infrastructure can and should be designed with this idea from the start. Standard, repeatable IT services can be established to enable rapid rebuilding:
Standard Operating Environments and configurations are established as a universal image for server and platform environments. Once the standard is set, new environments can be configured and distributed in minutes.
Personal Computers are gone and replaced by Production Computers. As such, the Production Computer is a well-defined, standardized tool with fixed components and applications. Installs and re-installs simply and without drama.
Applications on our servers and Production Computers are steered to a standard profile configuration with minimal variation. Configure and install like an app on your phone, there is no excuse for using older or unapproved versions.
Company-owned tablets and phones can be configured and deployed with the company's apps, contact book, and standard setup. Manage them thereafter as a standard item within a Mobile Device Manager tool.
DevOps has made this discussion popular with the “pets versus cattle” comparison.* “It used to be that when servers got sick, you nursed them back to health like a pet. In today’s standardized cloud, servers are numbered, like cattle in a herd. When a server gets sick, it’s taken out back and shot, replaced by next in line.”
Perhaps cruel to the cattle, but our IT services don’t have feelings.** What are the benefits?
Deploy new capacity fast? Install and configure in minutes because it's all pre-defined.
Incident repair that exceeds 15 minutes? Rebuild and reboot rather than wasting more time.
Version control? Piece of cake because it's easy to build and deploy everything to the same standard.
Security risk? Kill the problematic server or phone or pc and rebuild.
It really becomes an affirmation of Lean IT and DevOps appeals to standardization. And our challenge becomes the discipline of establishing and maintaining standards. Focus on what is needed, eliminate what is not, and solidify your standards. If we can, IT services are easily distributable through rebuild. When that occurs rebuild becomes easier than repair and the speed and accuracy of our service delivery improves.
There may be some final rhyme to my story about cattle and summer grilling, but I think I’ll heed my own advice. Rather than fixing this blog post any more, I’ll just kill it now…
*Attributed to Bill Baker, Microsoft.
** Really? Let’s get into this some other time.