11 comments

[ 4.1 ms ] story [ 32.1 ms ] thread
I think we are in front of a new trend. Moving out the cloud and back to bare metal servers. Will we see a decline of AWS in favor of Digital Ocean, Linode, Hetzer and Cloud66 ?
You're proposing cloud to...cloud?

I think the point is going much closer to self-hosted bare metal in colocation—fewer managed services with dubious value-adds.

Except for companies who do not want to deal with the infrastructure management. The last thing I want to do is deal with disk failures etc.
Very interesting set of design decisions. Really does seem so much simpler. Not sure how exactly they handled some of the tricky bits of state (such as disk volumes and environment variables) but it’s at least “working” in the video.
I like that these folks don't just brag "moving off of the cloud" but also sharing their tools to the others. This helps a lot of people save time and money. I am thinking .. that's it? What's missing? I'm on DO app platform now, will try to use their VPS instead with this MRSK.
FYI this is not a competitor to k8s. It is basically scp for docker containers. It doesn’t manage your running containers. You will need to hire people todo that with MRSK. Which is a choice and a trade-off. Not sure if it makes sense when big enough.

For example: - no auto scaling, auto container placement across nodes, therefore less efficiënt and more expensive in terms of resources - not handling node failure, no redistribution of containers - single node cron: if down, cronjobs stop working

K8S is a bit complex. But it solves a complex problem. MRSK only solves the deployment part.

It’s like saying a spaceship is too complex so we built a plane. Fine, but it doesn’t bring you to space….

That said, most people don’t need a spaceship. A plane will work just fine.
You could do the same with a easier UI by using Cloud66
Is it time for cloud stock rebalancing ?

- Sell Amazon

- Buy Digital Ocean, Dell