Someone has DM'd me asking if they should choose @RedHat@openshift on @awscloud (the fact that three Twitter accounts comprise the service name aside).
Oh dear me.
"Are you a 20 year old company with an existing investment in OpenShift?"
"No, we're a young startup."
"Then no, you should absolutely not use ROSA to manage your containerized workloads."
Seriously, look at the ROSA marketing pages. It talks about seamless migration to cloud from on-premises (always a lie) combined with the familiarity of the OpenShift workflows you're used to for your existing staff.
If you're building something greenfield, why would you go down the OpenShift rabbit hole in a pure cloud environment?
"Well, we hired a bunch of CNCF zealots."
So your question is whether selecting the wrong platform will be mitigated by also hiring the wrong people?
Half of the AWS marketing page for ROSA is taken up by the partners who'll run it for you.
So you want to orchestrate containers purely in AWS, and to do this you want to partner with AWS, RedHat, and some third party as well to do that?
Please don't do this.
I'm not saying ROSA is a bad service. I'm saying that it's aimed at big-E Enterprises with existing investments in OpenShift. (If I'm wrong, then its marketing needs a rework.)
The right answer here is ECS, specifically Fargate if possible. If that falls down, EKS.
But this is the most expensive container offering AWS has, with arguably some of the worst DX (that debate would be fierce!).
It serves a purpose and has a place, but it's not aligned at workloads that are brand new.
If you're hell bent on running OpenShift in AWS, yes; use ROSA. Managed services invariably beat Ruin Your Own Adventure stories.
But YIKES.
Another DM has come in. "We're migrating to AWS at (FORTUNE 100). We're using ROSA because of some corp dev deal, and it's been an unmitigated shitshow."
I don't blame customers for this, though; I blame AWS Marketing.
With this many different container offerings that are all remarkably similar, which should a customer pick? There *is no consistent guidance.*
The fact that this thread is *useful* to people is an example!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Back up any personal (NOTE: NOT CORPORATE IP!) data on my work laptop whenever I get a context-less "let's talk" message.
Putting all of my corporate expenses on my personal card, then expensing them instead of the other way around to avoid giving them the "well technically this might be embezzlement" stick if they disagree with a decision.
It wants an S3 bucket. Cool. The service has one bucket for ingest, and one for output generally. Which does this form want?
ALWAYS WITH THE QUESTIONS, YOU PEOPLE!
"Encryption of S3 buckets is basically a box check for compliance groups. That said, we at @awscloud are going to mandate it for this service with your own KMS key because the KMS team bribed us for that sweet $1 a month revenue juicer."
So back in 2008-ish, I volunteered as freenode network staff, which I was for the next seven years. This still occasionally surprises people. My nick was, unsurprisingly, "Corey."
I'd been on the network learning for a while, primarily in the postfix and centos channels. (That's how I met @BitIntegrity, whom I later married. To someone else.)
It was nice to have a chance to give back.
There was a lot of fun, a lot of drama... @bequinning and I flew out for annual holiday gatherings in the UK for a number of years.
Eventually I stopped having time to spend on IRC, and focused on other things. Like shitposting here!