Interesting Ping Problem with Gigaman Circuits and Cisco 4006 Switches - Anyone?

Discussion in 'Cisco' started by Cpt_Cam, Sep 27, 2004.

  1. Cpt_Cam

    Cpt_Cam Guest

    I have a customer that complains of slow access time accross a WAN
    circuit. The circuit itself is a Gigaman (1000MB Fiber). We're
    basically extending layer 2 out to them. At the main facility we have
    a 6513 with a L3 vlan interface and at the remote end we have a 4006
    configured in the same VLAN. GBIC's at both sides. The phone compnay
    tested the line and no errors were reported on their end. On the Cisco
    devices, all interfaces show clean with no errors (layer 2 and layer
    3).The odd thing is, by setting up steady pings to the IP of the
    switch I get sporadic response times ranging from 6ms all the way up
    to 300ms. If I ping computers behind the switch in the same VLAN, I
    get steady 1ms response times. Anybody can explain that anomolie?

    As a sidenote, we have other gigaman circuits with 4006's connected to
    them, same IOS (5.5(10.0))ver bootstrap (5.4(1) and they ping steady
    6-12ms. Devices behind them also exhibit this behavior. (where they
    ping steady at 1ms).

    Boggled.
    Cpt_Cam, Sep 27, 2004
    #1
    1. Advertising

  2. Cpt_Cam

    Ivan Ostreš Guest

    In article <>,
    says...
    > I have a customer that complains of slow access time accross a WAN
    > circuit. The circuit itself is a Gigaman (1000MB Fiber). We're
    > basically extending layer 2 out to them. At the main facility we have
    > a 6513 with a L3 vlan interface and at the remote end we have a 4006
    > configured in the same VLAN. GBIC's at both sides. The phone compnay
    > tested the line and no errors were reported on their end. On the Cisco
    > devices, all interfaces show clean with no errors (layer 2 and layer
    > 3).The odd thing is, by setting up steady pings to the IP of the
    > switch I get sporadic response times ranging from 6ms all the way up
    > to 300ms. If I ping computers behind the switch in the same VLAN, I
    > get steady 1ms response times. Anybody can explain that anomolie?
    >
    >


    How long the fiber is? Which GBIC's are you using? Which fiber are you
    using?

    --
    -Ivan.

    *** Use Rot13 to see my eMail address ***
    Ivan Ostreš, Sep 27, 2004
    #2
    1. Advertising

  3. Hello, Cpt_Cam!
    You wrote on 27 Sep 2004 09:25:14 -0700:
    CC> clean with no errors (layer 2 and layer 3).The odd thing is, by
    CC> setting up steady pings to the IP of the switch I get sporadic
    CC> response times ranging from 6ms all the way up to 300ms. If I
    CC> ping computers behind the switch in the same VLAN, I get steady
    CC> 1ms response times. Anybody can explain that anomolie?

    What's CPU load on 4006? Responding to ICMP packets is among the last things
    switch will do.

    CC> As a sidenote, we have other gigaman circuits with 4006's
    CC> connected to them, same IOS (5.5(10.0))ver bootstrap (5.4(1) and
    CC> they ping steady 6-12ms. Devices behind them also exhibit this
    CC> behavior. (where they ping steady at 1ms).

    And what's CPU load on those.

    With best regards,
    Andrey.
    Andrey Tarasov, Sep 28, 2004
    #3
  4. Cpt_Cam

    Cpt_Cam Guest

    > What's CPU load on 4006? Responding to ICMP packets is among the last things
    > switch will do.


    CPU Load = 5 Min @ 52%, One Minute @ 52%, 5 seconds @ 55%
    ~26-31% of the utilization is in a category called "Switching
    Overhead"

    >
    > CC> As a sidenote, we have other gigaman circuits with 4006's
    > CC> connected to them, same IOS (5.5(10.0))ver bootstrap (5.4(1) and
    > CC> they ping steady 6-12ms. Devices behind them also exhibit this
    > CC> behavior. (where they ping steady at 1ms).
    >
    > And what's CPU load on those.


    Low Utilization is 11% - Highest is 17%



    How long the fiber is? Which GBIC's are you using? Which fiber are you
    using?

    Multimode fiber from the switches to the Phone Companies device
    (extender or whatever they want to call it). These are essentially
    fiber patch cables no longer than 15 feet. Then, couldn't tell you
    what fiber they're using to extend out over 6 miles. I would assume
    single mode fiber. How can I verify what GBIC is in the switch
    remotely using the CATos? I usually draw them from a supply of
    WS-G5484's but I can't get to the switch to tell for sure.

    Thanks for the help thus far.
    Cpt_Cam, Sep 28, 2004
    #4
  5. Hello, Cpt_Cam!
    You wrote on 28 Sep 2004 15:37:27 -0700:

    ??>> What's CPU load on 4006? Responding to ICMP packets is among the
    ??>> last things switch will do.

    CC> CPU Load = 5 Min @ 52%, One Minute @ 52%, 5 seconds @ 55%
    CC> ~26-31% of the utilization is in a category called "Switching
    CC> Overhead"

    ??>> And what's CPU load on those.

    CC> Low Utilization is 11% - Highest is 17%

    Well, there is a difference, isn't it? From

    http://www.cisco.com/en/US/products/hw/switches/ps663/products_tech_note09186a0080094956.shtml

    Switching Overhead is actually a process consisting of several subprocesses that
    handle the following:

    Address learning (path setup) for new MAC addresses

    Normal host entry aging, as well as fast-aging, due to reception of STP Topology
    Change Notification (TCN) Bridge Protocol Data Units (BPDUs)

    Packet processing for control traffic, such as STP BPDUs, CDP, VTP, DTP, and
    PAgP

    Packet processing for management traffic, such as Telnet, SNMP, and HTTP, as
    well as broadcast and multicast packets in the sc0 or me1 subnets

    So the question is what's hammering the box with 50% utilization?

    CC> Multimode fiber from the switches to the Phone Companies device
    CC> (extender or whatever they want to call it).

    We had JDS Uniphase box. I think this is it -

    http://www.jdsu.com/site/images/products/pdf/Model1280GbX_Gigabit_EthernetExtender.pdf

    There is nothing which could affect RTT without affecting traffic.

    With best regards,
    Andrey.
    Andrey Tarasov, Sep 29, 2004
    #5
  6. Cpt_Cam

    fearo Guest

    On Tue, 28 Sep 2004 17:19:20 -0700, Andrey Tarasov wrote:

    > Hello, Cpt_Cam!
    > You wrote on 28 Sep 2004 15:37:27 -0700:
    >
    > ??>> What's CPU load on 4006? Responding to ICMP packets is among the
    > ??>> last things switch will do.
    >
    > CC> CPU Load = 5 Min @ 52%, One Minute @ 52%, 5 seconds @ 55%
    > CC> ~26-31% of the utilization is in a category called "Switching
    > CC> Overhead"
    >
    > ??>> And what's CPU load on those.
    >
    > CC> Low Utilization is 11% - Highest is 17%
    >
    > Well, there is a difference, isn't it? From
    >
    > http://www.cisco.com/en/US/products/hw/switches/ps663/products_tech_note09186a0080094956.shtml
    >
    > Switching Overhead is actually a process consisting of several subprocesses that
    > handle the following:
    >
    > Address learning (path setup) for new MAC addresses
    >
    > Normal host entry aging, as well as fast-aging, due to reception of STP Topology
    > Change Notification (TCN) Bridge Protocol Data Units (BPDUs)
    >
    > Packet processing for control traffic, such as STP BPDUs, CDP, VTP, DTP, and
    > PAgP
    >
    > Packet processing for management traffic, such as Telnet, SNMP, and HTTP, as
    > well as broadcast and multicast packets in the sc0 or me1 subnets
    >
    > So the question is what's hammering the box with 50% utilization?


    Our network consists entirely of Catalyst 4006/6500 layer2 devices. We
    experienced this exact same issue with "switching overhead" consuming
    large quantities of cpu time.

    Here are some actions that we have taken in our switching environment to
    reduce our cpu load from 75-85% to just 20% which is typical.

    Recommended Actions:

    1) reduce the total amount of spanning tree instances within your vtp
    domain.

    a) groom your existing vtp domains to remove any excess vlan created in
    error or that are currently unused.
    b) Use Routers to interconnect sites via layer3 address splitting your vtp
    domains into smaller areas.
    c) *extreme measure* switch from per vlan spanning tree (pvst+) to
    multiple instance spanning tree "mist" allowing you to bundle your vlans
    together reducing your cpu load. All devices will need to be mist
    compliant.

    2) remove management device interfaces from client networks by create a
    unique administrative vlan and associated layer3 network that restricts
    access to admins only.

    3) prevent ordinary users or servers connecting to access ports sending bpdu
    information into your switching environment by implementing
    bpdu-gaurd or bpdu-filter on non switch to switch ports. This will
    prevent any unwarranted spanning tree topology changes.
    fearo, Sep 29, 2004
    #6
  7. Hello, fearo!
    You wrote on Wed, 29 Sep 2004 04:38:14 GMT:

    f> 3) prevent ordinary users or servers connecting to access ports
    f> sending bpdu information into your switching environment by
    f> implementing bpdu-gaurd or bpdu-filter on non switch to switch
    f> ports. This will prevent any unwarranted spanning tree topology
    f> changes.

    I wouldn't use bpdu-filter unless there is no other choice.

    With best regards,
    Andrey.
    Andrey Tarasov, Sep 29, 2004
    #7
  8. Cpt_Cam

    Hansang Bae Guest

    In article <>,
    says...
    > I have a customer that complains of slow access time accross a WAN
    > circuit. The circuit itself is a Gigaman (1000MB Fiber). We're
    > basically extending layer 2 out to them. At the main facility we have
    > a 6513 with a L3 vlan interface and at the remote end we have a 4006
    > configured in the same VLAN. GBIC's at both sides. The phone compnay
    > tested the line and no errors were reported on their end. On the Cisco
    > devices, all interfaces show clean with no errors (layer 2 and layer
    > 3).The odd thing is, by setting up steady pings to the IP of the
    > switch I get sporadic response times ranging from 6ms all the way up
    > to 300ms. If I ping computers behind the switch in the same VLAN, I
    > get steady 1ms response times. Anybody can explain that anomolie?
    >
    > As a sidenote, we have other gigaman circuits with 4006's connected to
    > them, same IOS (5.5(10.0))ver bootstrap (5.4(1) and they ping steady
    > 6-12ms. Devices behind them also exhibit this behavior. (where they
    > ping steady at 1ms).



    Packets that terminate at the router are process switched. So that
    could be it.


    --

    hsb

    "Somehow I imagined this experience would be more rewarding" Calvin
    *************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
    ********************************************************************
    Due to the volume of email that I receive, I may not not be able to
    reply to emails sent to my account. Please post a followup instead.
    ********************************************************************
    Hansang Bae, Oct 5, 2004
    #8
    1. Advertising

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Erik DLS

    Gigaman Light Levels

    Erik DLS, Dec 11, 2003, in forum: Cisco
    Replies:
    1
    Views:
    1,159
    Andrey Tarasov
    Dec 11, 2003
  2. Rutger Blom
    Replies:
    0
    Views:
    503
    Rutger Blom
    Jun 4, 2004
  3. Catalyst_user

    TFTP from a Cisco Catalyst 4006 problem

    Catalyst_user, Mar 17, 2005, in forum: Cisco
    Replies:
    5
    Views:
    773
    Hansang Bae
    Mar 22, 2005
  4. Replies:
    1
    Views:
    841
    donjohnston
    Dec 30, 2008
  5. Greg
    Replies:
    5
    Views:
    8,753
    Sarcasmus
    Jul 1, 2013
Loading...

Share This Page