Technology
This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
view the rest of the comments
I hear ya.
I'm an old school linux sysadmin. Been at it since 2003. I always groan when some higher-up wants us to host stuff on AWS and kubernetes, because they have no idea how stuff works, they just like the buzzwords and how everybody else is doing it this way.
And I'm just thinking, great. Instead of renting a very affordable beefy server with Hetzner or OVH, we instead use AWS which costs us 3-5x as much for less performance. And I try to explain to them, that not every application is like Netflix where you can just throw more nodes at it in order to scale.
Docker I like, since it can make things easier, with regards to ensuring a consistent environment, once you know your way around it. But kubernetes and AWS I hate with a passion. With AWS I can never truly know how much it's gonna cost in advance, since their pricing structure is so mind boggingly complicated and you get billed even if so much as a fly farts in one of their server rooms.
Another thing is, every server provider that I've ever worked with, provides you with a means to boot a rescue mode if something goes wrong, so that you can fix it and reboot. No, with AWS, you need to make a snapshot, fire up another instance, attach that snapshot as a volume and mount it, then try and fix stuff, then shutdown that instance, detach that volume, then re-attach it using the same device name as the root device on your original instance. Oh, doesn't boot? Rinse/Repeat. This shit can take hours. Whereas with a "normal" root server and a IPMI console and a rescue boot, it's a question of minutes (usually).
Same with backups. I always ensure to create a backup system, so that I can restore certain files individually. Yeah, you can have automated snapshots of your EC2 Instance. But guess what, if something goes wrong, you boot the entire snapshot. So even if you just need to recover a single file, you still need to go through the process of reverting the entire system to a previous snapshot.
Bah, I could go on and on, but honestly the whole thing just seems to create more and more abstraction layers that just makes it difficult to figure stuff out when something doesn't work.
The only thing, that I can somewhat appreciate with AWS is the ease with which you can migrate to a more powerful instance if you need it. Or how quickly you can launch a new instance. But thats about it.
In previous companies, we would have a couple of beefy servers that would run multiple apps on a single server, with perhaps a failover server being at the ready. Our server Budget was maybe 1-2k € per month.
With AWS it's like 5x that.
You just need Terraform. 🤭
On a serious note, Docker is brilliant for many reasons, some of which you mentioned. I think K8s is great too. With that said, running K8s just because is mindbogglingly stupid. I've seen this done in corpos I've worked exactly as you said it. In addition, one doesn't have to run K8s on AWS. Or any other public cloud. Or anywhere where it doesn't make sense. You make pretty good points about AWS. If it's better in some way to run on metal, you run on metal. And if it makes sense for an application to be deployed on K8s, that might also be run on top of the metal you have. But here's another and I think potentially more significant point against AWS and the likes. They introduce lock-in. Every public cloud has its own services with their own APIs and features. The moment you buy into those, e.g. Cloudwatch automation, you can no longer move the workload to a place that doesn't have Cloudwatch. And no place other than AWS has it. And you're locked in. 👏 Developers (me) find 10 easy solutions in some marketing wank from AWS, all of them proprietary, stitch them together to implement some service and bam, that service forever belongs to AWS. You pay per fly fart from here on out, until you rewrite it.
In fairness to Terraform, it does indeed make life easier if you have to juggle workloads across public clouds.