More than one 1G nic, more than one unmanaged switch?

Discussion in 'Server Networking' started by Boe, Jan 20, 2006.

  1. Boe

    Boe Guest

    I'm not familiar with using more than one NIC. I have two netgear
    unmanaged gigabit switches that connect via a single crossover cable. Would
    I be better off putting all of my heaviest users (transferring the largest
    files all with gigabit nics on their systems) on one switch and putting both
    of the server nic connections (where all the traffic is coming and going to)
    on a single switch or should I put half on one switch and half on another
    switch (heavy users) and put one nic connection from the server on each
    swtich? (we don't have funds to replace the switches so that will not be
    part of the equation)

    Also since I am a newbie can you give me details on anything I would need to
    do on the server once I implement whatever solution is better. The only
    server connection plugged in right now is one nic at 10.10.10.10. I'm not
    sure what I'll need to do with the second nic or any loadbalancing things I
    will or won't need to do.

    Thanks!
     
    Boe, Jan 20, 2006
    #1
    1. Advertisements

  2. Boe

    Boe Guest

    Yes, the clients are frequently transferring files of 500MB or larger
    simultaneously.
     
    Boe, Jan 21, 2006
    #2
    1. Advertisements

  3. Boe

    Boe Guest

    I thought YES answered the question. What was the question I didn't
    answer?

     
    Boe, Jan 21, 2006
    #3
  4. Boe

    Boe Guest

    The question has changed. I found out that at least half of the clients who
    have large file transfers are on a switch that is not in the server room.
    Now what is the best way to optimize performance as the Nic on the server is
    maxing out - checked using task manager - network monitoring.

     
    Boe, Jan 21, 2006
    #4
  5. Boe

    Boe Guest

    A single G5 Macintosh client exceeds 50% with large file transfers. We have
    over 30 developers using them on our network.

     
    Boe, Jan 21, 2006
    #5
  6. Boe

    Boe Guest

    Do you need to see my taxes? Clearly you have no interest in answering this
    question. If your questions actually were leading to a solution I would
    continue with this but clearly you are the helpful sort.

     
    Boe, Jan 21, 2006
    #6
  7. Boe

    Boe Guest

    I have stated by watching task manager I can see it max out. In the event
    it assists someone who wants to HELP me, I am running Windows 2003 with all
    the latest patches on a Dell 2800 server using the native nics on the system
    with the latest drivers.

     
    Boe, Jan 21, 2006
    #7
  8. Boe

    Bill Grant Guest

    My personal opinion is that putting multiple NICs in a machine causes
    more problems than it solves, unless you use NIC teaming.
     
    Bill Grant, Jan 21, 2006
    #8
  9. Boe

    Boe Guest

    Thank you for your honest opinion.

     
    Boe, Jan 21, 2006
    #9
  10. Boe

    Bill Grant Guest

    I agree with CJT that you really need to establish what is actually
    happening on the network. If you do establish that the server NIC is the
    bottleneck you could look at using two teamed NICs. It won't double the
    throughput but should give you quite a boost. Have you talked to Dell about
    this?
     
    Bill Grant, Jan 21, 2006
    #10
  11. Boe

    Boe Guest

    I'm trying to identify what the issue is. We are using EztremeZ-IP for our
    macs because native SMB or AFP was unbearable. We get the fast throughput
    most of the time (rarely when we didn't use it) but we do max out the NIC on
    the server occassionally and for some reason it takes a while to recover.
    The disk drives are fine - not too much activity but it takes about 15
    minutes before performance goes back to normal.

     
    Boe, Jan 22, 2006
    #11
  12. Boe

    Boe Guest

    No I haven't. I've never asked Dell Windows questions before.
     
    Boe, Jan 22, 2006
    #12
  13. Boe

    Boe Guest

    There is very little I can give but if someone asks a question I can answer
    I'll be happy to do so.
    There are NO events in the event viewer at the times of the slow downs.
    The disk array activity does not spike.
    The only thing I do see is that the network card (using latest driver -
    update monthly - no improvement) does max out.
    CPU never exceeds 25%
    Memory - 2 GIG - never exceeds 1 GIG
     
    Boe, Jan 22, 2006
    #13
  14. There is very little I can give but if someone asks a question I can
    What Bor & CJT are trying to tell you is that, at the begining, you must
    find the "botleneck" in your LAN. So try to use the Perfomance Monitor winth
    Output Queue Length counter to be sure, that your server's NIC indeed is the
    "troble maker".
     
    Iuri Cuznetov, Jan 23, 2006
    #14
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.