This message was deleted.

No? You can mix amd64 and arm64 if you want. Unless the workload cares for some reason? We don't care.

Doing this presently across x86 and ARM...no impact, except as mentioned, one of our workloads is not ARM friendly, so we "taint" around it.

Ok - so what I talk about is using Intel Xeon and AMD Epyc which are both x86 CPUs. So they are both using the same architecture. I don´t have any concerns about this, but the user side is sceptical.

What you are saying is correct. There is no user concern to be had, when using AMD and Intel; in your scenario.

Thanks Bryan. Great to hear! 🙂

What specifically are they sceptical about? The only thing I've ever seen care about specific professor models is VMs, and even in that case it just limits live migration.

There is some minor Platform difference among x86 with regards to some extensions used in GPU/Crypto Stuff, e.g. Intel QuickSync or IntelQAT, which AMD doesn't have equivalence to...but like @creamy-pencil-82913 was saying, its common to VMs or EXTREME Baremetal Use cases like a HW Firewall, Cross Domain Solutions, Hardware Security Modules in Cert Authorities, etc. I've not seen any of those in Container Workloads yet.

The users are sceptical around the facts that Bryan mentioned in the post before. At the moment, nobody can justify it in a meaningful way or show a workload that would be affected by it.