This month’s T-SQL Tuesday invite comes from Koen Verbeeck, who asks us to think about what it would look like to move back to on-premises infrastructure after years in the cloud. Do we still have the skills to go back to on-premises? What types of on-premises issues would be the most frustrating?
Click the T-SQL Tuesday logo to read the full invite.
The Server Room
When I think of on-premises servers, nostalgia kicks in and my thoughts go straight to “the server room” from my college years. The server room was sacred. Very few people could enter, as it required an access code. The few times I was in it gave me the same feeling as being in a cramped antique store where you’re afraid to touch anything or accidentally knock over an expensive lamp.
Back then, when hardware failed, it was a matter of ordering parts, waiting for them to come in and, if it was for a server, going into that warm room to make the change. Yes, kids, we had to use our hands. And thanks to the frustrating design of Gateway cases on the desktop PCs on campus, many cuts were had on those hands. But I digress.
These days, in the cloud, a hardware issue can be as simple as a few clicks, and you move on with your day.
I wish I knew what that server room was like today and how many servers are left. Even if there aren’t many servers there today, there might be tomorrow, as the trend may be leaning towards moving from the cloud back to on-premises.
Not So Fast
As I alluded to before, it used to be a more time consuming hassle to deal with hardware failures and upgrades. In the cloud, this type of work can be handled at the click of a button if it’s not handled automatically. We’ve been spoiled by how easy it can be. Depending on what hardware you need, availability is another issue that would need to be considered. Maybe that’s an edge case, and ideally wouldn’t be an issue to face often, but it’s one I could see being frustrating to deal with.
From a database administrator perspective, moving back to on-premises could also mean more work optimizing queries. The “solution” would no longer be as quick to throw more memory or other resources at a problem and carry on. Rather than adding band-aids and tech debt, there would be more effort towards fixing the root cause of problems.
The future seems to be somewhere in between the cloud and back to on-premises. I don’t think that’s a step backward or a failure, just an adjustment to the world we live in.
Thanks for reading!
Chad Callihan 