Question for the experts:
We recently purchased an updated AMD to replace retiring hardware but the supported 10GB card was not delivered and it might take some time to get this resolved.
I'm anxious to press the new server into service because we are hitting the ~2GB throughput ceiling on the older classic device. Should I wait for the correct supported 10GB card to use the customized drivers, or will HS AMD at 5GB rate work ok with Native drivers? Other server stats are 128GB ram, 16 cores, etc. (plenty of capacity).
Any insights or suggestions would be appreciated.
Personally, I would expect it to work okay with the native drivers at 5 Gb, but as always, it will depend on your specific traffic and analysis. You can always test and if it doesn't handle with the native drivers, wait for the custom driver supported card to arrive and put it into service then.
After running for a couple of weeks on AMD classic with native drivers there have been a few issues with core dumps on the server. Dev has suggested that the packet rate is too high (300-900kpps) for the native drivers to handle and custom drivers might help keep from running out of memory. I'm curious if the architecture/increased throughput capacity in HS AMD would help alleviate this, or compound the issue? Probably not knowable at this point, but figured I'd throw it out there for comment.
Thanks Erik - That's my gut as well, glad to hear there isn't any glaring issue to be concerned about. The old one is constantly sampling now, how much worse can it be? I'll see how it goes.