Comment on Europe is looking for alternatives to US cloud providers
aleq@lemmy.world 3 weeks ago
I sure hope so, but I have little faith tbh. Cloud providers have done a great job selling serverless solutions that are tightly coupled with the provider. Wise companies have limited themselves to the basics - load balancers, servers, maybe some serverless container solution or kubernetes. The latter can move pretty much anywhere with some, but not a whole lot, of effort. The former, have fun rediscovering the quirks of your new provider’s equivalent of lambdas or whatever (or at worst, rewriting the whole thing).
partial_accumen@lemmy.world 3 weeks ago
“Wise” is subjective here. Using a cloud vendor’s implementation can yield many times more efficiency, simplicity, stability, scalability, and agility vs rolling you own. Does it come with the cost of vendor lock-in? It absolutely can. Will that make migration to another vendor difficult? It will.
So for organizations that never embraced the cloud alternatives have had to maintain their own infrastructure or use commodity solutions, as you mentioned, to deliver their IT needs. How much more was spent using a general purpose approach with higher portability to deliver the same result vs a cloud providers proprietary version? Then include the time component.
Only time will tell.
ubergeek@lemmy.today 3 weeks ago
So far, speaking from experience, we saved loads of money DIY’ing it, even when deploying to the cloud, and we saved loads of time, in the long run.
WE KNOW where the perf problem is. WE KNOW the cadence for how long a fix will take. WE KNOW the OS we’re deploying. WE KNOW the apps we’re deploying. WE KNOW how to squeak additional perf from the infrastructure, tuned to OUR NEEDS.
If you hire talented workers, you save money and time, by DIYing the approach, as long as it’s done in a sane, and controlled manner.
partial_accumen@lemmy.world 3 weeks ago
First, I’m glad its working for you up to now. I’ve been in similar orgs. It works great, until it doesn’t. Have you had an production outage yet from a datacenter or hardware failure yet?
Should I ask home much did your Broadcom licensing renewal cost you this year?
Talented workers that know the systems are great, and if you’ve built your own systems and processes finely tuned to your specific applications performance needs and profiles, it also means you’ve got a highly specialized infrastructure and app stack. You’ve possibly built yourself a scaling problem because the skill needed to understand and maintain your well performing one-off solution isn’t ubiquitous. As your organization’s needs scale it will be tied directly to the additional limited specialized and expensive staff needed. Again, this may not be an issue with your org today, but it may not have hit this need yet. This is the “Only time will tell” component that is so important. As in, your sample size may not be large enough to know if your org made the right decision or not yet.
ubergeek@lemmy.today 3 weeks ago
Yes, we’ve had outages in DCs. They are usually just a blip, because we have more than one.
And we don’t pay broadcom anything. We migrated off of esx a long time ago.
And the skills needed? We use a floss stack, so you need to know stuff like nginx, puppet, mariadb, and php.
Not exactly cutting edge stuff there.
Operations engineers make sure the infrastructure is up, and ready for code. Devs own the code.
So, no, it’s really not all that niche.