Why We are not getting 10Gb/s on 10Gb HP NC522SFP NIC and 10Gb Cisco 4900M switches?
We have 4 HP Proliant 580 G5 node cluster running Orcle Enterprise Linux 5.4 2.6.18-164.el5 x86_64
With interconnect of 2 10Gb Cisco 4900M switches and 10Gb HP NC522SFP NICs.
We are using IP Bonding with mode 1 active-backup mode with jumbo frames (8000 MTU).
8GB/s raw data - no processing at all. With 8k Jumbo frames. And a lot of packet receive errors. Not being a network specialist, I'd still recommend to drop the jumbo frames, and return to standard 1500 byte frames. Get rid of these packet errors.
What does this have to do with the product Oracle Application Interconnect?