Getting our applications out of the cloud provided the main celebration for our exit, but seeing the actual spend tumble is the prize. See, the only way to get pricing in the cloud down from obscene to merely offensive is through reserved instances. This is where you sign up for a year or more in advance on a certain level of spend. Th...
That’s the thing, ‘cloud’ is just another tool in your toolbox. It’s the right tool for some workloads and the wrong one for others. The fact they’ve shifted the work to their own servers and kept the ops team suggests it was the wrong sort of workload to be in the cloud in the first place.
For a while there was an obsession with moving everything to the cloud, and that was always going to be an expensive mistake in a number of different ways. Hopefully, as the hype dies down more nuanced decisions will be made. There’s a whole gamut of options between all in the cloud and all in the data centre, and when people jump straight from one end to the other I’m put in mind of Hamlet’s quote “There are more things in heaven and earth, Horatio, / Than are dreamt of in your philosophy.” Understand your workload, understand your business’ future plans and their needs, and then make a plan, considering all the tools at your disposal.
I hate the obsession to move to the cloud and the obsession towards serverless or functions.
Functions are stupid and crazy for anything that is actually used often.
For small utilities, they make a ton of sense, but next time I see an app with millions of requests per day using functions, I’m going to lose my mind.
Years ago I was the senior techie in designing and implementing distributed high performance server systems and what you reminded me of just made my blood start to boil… :/
Amazon is actually quite smart. They give away tons of free compute in the form of free tier lambda functions, enough to convince any hobbyist to comfortably port all of the home built stuff to the cloud for free, but not enough that any serious business can run cheaply.
Then they wait for someone to suggest using the tool they use at home for the company, and rake in the rewards.
It’s pretty crazy how “millions of requests per day” is somehow enough to warrant paying the Amazon price for cloud stuff. A $50 VPS can handle millions of requests per day with capacity to spare. If you’re in a real pinch, get a $100 dedicated server somewhere.
Computers are crazy fast. If you cut out all of these layers of abstractions, you’ll be surprised how little hardware you actually need to run the equivalent of cloud stuff. You’re not done for that cheap (you need to set up failovers, backups, and so on) but with the ridiculous prices Amazon and Azure charge, it’s worth running the numbers.
The only problem with that approach is that your product will slow down if you suddenly get an influx of millions of customers, something that never happens to 99% of companies considering cloud stuff.
If there’s anything that 3 decades in Tech have taught me is that fad-following commonly rules it, even with the supposedly logical (but not really) techies.
Cloud storage and cloud computing became a fad about a decade ago (I still remember the hype repeated by people who had never actually designed distruted systems) so there were tons of people jumping headfirst without a plan into it for the hype and the seemingly cheaper price (if you didn’t think your needs and future evolution through) even though it wasn’t the best choice for them.
No doubt well see the same kind of fad-following over making-sense-for-us thing with the latest hype-train: AI.