I have a situation where when I transfer the same file bewteen two offices it takes three times longer in one direction when compared to the other direction. So in the slow direction Office A to Office B in my traces I can see an excessive number of duplicate acknowlegements occuring at different times of the tarnsfer. I'm guessing this is causing my transfers to run slower in that direction. If anything I was expecting to see re-try requests (i.e. the packets not making it to Office B) as I have an interface in the path where I see output discards which may be due to NetApp microburts so I'm perplexed as to why I'm seeing so may duplicate acknowledgments coming back. The packet has obvously made it to the destingation so why so many acknowledgements. When I say excessive I typically see 200 duplicates acknowledging the same packet. asked 04 Nov '13, 00:39  MPN |
One Answer:
hard to tell without a capture file. There are a lot of possible reasons. Based on your description (difference between the direction of data transfer), it could be some kind of QoS solution that tries to hold back the ACKs, to enforce a slowdown of the transfer in one direction. Regards answered 04 Nov '13, 05:35  Kurt Knochner ♦ |
ok will take a look through the path.
Thanks Mark
what kind of connection do you have between the offices? MPLS, Ethernet, etc.?