Help with xmit/recv discrepancy

[JeffDPSU]

Hi - New to the board - long time lurker. Looking for some guidence, or at least the right direction to look. 

My backup server is on port 12 of Switch 4210G 48 port. This switch is on a 10 GbE uplink to a Procurve 2910al. I'm backing up data from port 9 on this switch. I expect that xmit traffic on the backup server to be much less in BPS than recv traffic. 

I monitor my LAN with Solarwinds Orion NPM 10.5.

Traffic statistics for these two ports while that job is running seem to be reported in reverse. Average xmit BPS on the server being backed up is 976 Kbps, average recv BPS on the same server is 308 Mbps

Average recv BPS on the backup server (CA ARCServe Backup r16 and BE 2012) is 2.3 Mbps while average xmit BPS on the same server is 633 Mbps.

First thing I did was contact Solarwinds support - they had me run snmpwalk before a backup job using the same interfaces and another one after. They analyzed the results and determined that the numbers that were being collected by the application were the same as the numbers being reported by the switch. In other words, it's not their product reporting the numbers incorrectly, it appears to be the switch.

I've been a sysadmin for several years, but I've never had to troubleshoot anything like this, so I'm really stumped on what I should be looking for. 

I would appreciate any suggestions.

Thanks

Jeff

[Vince_Whirlwind · Accepted solution]

I think that makes sense:

The server being backed up is sending data to the switch - the switch records this as RECV BPS.

The data makes its way through your network and when it arrives at its destination, it is being XMIT BPS out the port on the 4208.

[JeffDPSU]

Wow, that seems really simple. The xmit stat is for the interface transmitting to to the backup server. So the recv stats should match up with the xmit stats of the sending interface, regardless of the the xmit stats of the interface to the backup server. Thanks for the clarification.