Re: 100Mbps Ethernet Speed/Efficiency

From: Neall (neall_at_haughtmail.com)
Date: 04/06/04

  • Next message: Rodolfo J. Paiz: "Re: Sendmail with "more domains"" 
    Date: Tue, 6 Apr 2004 15:18:11 -0600 (MDT)
To: fedora-list@redhat.com

    >> I'm wondering what's the maximum sustained transfer rate
    >> that one can experience when using a 100Mbps link?

    I took the hard drives out of the picture. I set up RAM disks on two
    different systems, and connected them via ethernet using NFS to cross
    mount them. Then, I copied from the first system's RAM disk over
    ethernet, to the second system's RAM disk. Granted, NFS is in the
    picture, but it would be using hard drives, too.

    These numbers include NFS overhead (obviously):

    100 Mb file:
    100/T: 11.335 MByte/Sec (averaged over many tests).
    1000/T: 69.9 MByte/Sec.

    Obviously, the 1000/T (gigabit) test would have immediately been
    bottlenecked by the drive transfer rates. The 100/T rates are near
    theoretical. Not sure where the 1000/T bottle neck is, but I used stock
    NFS parameters. I may try to transfer via socket or rpc, but just haven't
    gotten there yet. I figured most people transfer via NFS (or SAMBA)
    mounted drives, so wanted to start there.

    Still, ~70 MByte/sec isn't shabby over gigabit ethenet via NFS. Neall 

    -- 
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: http://www.redhat.com/mailman/listinfo/fedora-list
  • Next message: Rodolfo J. Paiz: "Re: Sendmail with "more domains""

    Relevant Pages

    • Re: 100Mbps Ethernet Speed/Efficiency
      ... I copied from the first system's RAM disk over ... Granted, NFS is in the ... > picture, but it would be using hard drives, too. ...
      (Fedora)
    • Re: 100Mbps Ethernet Speed/Efficiency
      ... I copied from the first system's RAM disk over ... > ethernet, to the second system's RAM disk. ... > picture, but it would be using hard drives, too. ...
      (Fedora)
    • Re: NFS and kernel cache
      ... NFS that is your bottleneck and it is related to normal buffered ... of the hard drives, ... (still reading from the client hard drive with ntfsclone and writing ... the client (no network involved), the hard drive LED is on ...
      (Fedora)
    • Re: NFS and kernel cache
      ... caching writes to NFS (but I can't tell if it is the filesystem layer or ... NFS that is your bottleneck and it is related to normal buffered ... local hard drives. ...
      (Fedora)