Hi All, I was just wondering if someone can possibly explain some NFS situation to me? I've tested NFS over a WAN type environment in two scenarios, both of them, being horribly slow. Scenario one: -------------- Server is in London, on a 2MB leased line, while the client, was on a 128KB Leased line, running a vpn and blowfish encryption. bash-2.03# time dd if=/dev/zero of=/root/tmp/test-60M-1 bs=16k count=4096 4096+0 records in 4096+0 records out 67108864 bytes transferred in 56541.615297 secs (1187 bytes/sec) real 942m35.448s user 0m0.010s sys 0m1.930s MRTG showed that the 128K line was nowhere near peaking, and thus, can I be right to presume that NFS did not utilise all the available bandwidth to transfer this file? Scenario two: --------------- Server is in London, on a 2MB leased line, while the client was on a 540KB Leased line, without running over the encrypted VPN. su-2.03# time dd if=/dev/zero of=/root/tmp/test-60M-1 bs=16k count=4096 4096+0 records in 4096+0 records out 67108864 bytes transferred in 25887.387617 secs (2592 bytes/sec) real 431m48.407s user 0m0.017s sys 0m1.177s Once again, my MRTG graphs showed NFS did only utilise a fraction of the available bandwidth... If anyone know what I'm doing wrong, or how I can improve these figures, PLEASE let me know??? Surely, over 540KB Leased Lines, 2.5KB/Sec is just plain and simply ridiculous? We're resting between providers at the moment to get the one who will best suite our needs, but it's starting to look as if it's rather a software issue than a provider / network issue, so any help will be very much appreciated. Regards, Chris Knipe Technical Administrator Vardus (Pty) Ltd Cape Town - South Africa Tel: (+27) 21 670 9880 Fax: (+27) 21 674 4549 Cell: (+27) 83 430 8151 - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
