Catalyst 4500 Redundancy

Discussion in 'Cisco' started by PurpleServerMonkey, Aug 23, 2008.

  1. We currently have a single Cisco Catalyst 4510 with dual supervior IV
    modules and dual power supplied. It's central to our network with
    workgroup switches and servers connecting to it.

    My question is how internally redundant is a Cisco 4510?

    Basically investigating our options in making the core more fault
    tolerant but haven't found a lot of information around what could go
    wrong with a 4510. Trying to decide whether we should add another 4500
    series switch to the core or upgrade to something like a 6500.

    Thanks in advance for your feedback.
     
    PurpleServerMonkey, Aug 23, 2008
    #1
    1. Advertising

  2. PurpleServerMonkey

    Stephen Guest

    On Sat, 23 Aug 2008 05:31:50 -0700 (PDT), PurpleServerMonkey
    <> wrote:

    >We currently have a single Cisco Catalyst 4510 with dual supervior IV
    >modules and dual power supplied. It's central to our network with
    >workgroup switches and servers connecting to it.
    >
    >My question is how internally redundant is a Cisco 4510?
    >

    there are backplane traces to both Sups from each card, so you can
    tolerate loss of most things as long as your network design does not
    depend on a single I/O blade.

    >Basically investigating our options in making the core more fault
    >tolerant but haven't found a lot of information around what could go
    >wrong with a 4510. Trying to decide whether we should add another 4500
    >series switch to the core or upgrade to something like a 6500.


    The old joke is that it is not resilient until it could survive a
    nutter with a sledgehammer :) - but not many sites meet that spec...

    6500 gets you more of everything - bandwidth and flexibility mainly
    but you also get Netflow on some setups which might be importantm and
    better inservice upgrade - but 1 box has inherent limitations.

    so i would go for 2 4510s as dual star points for the network - even
    if they do not have dual sups.

    1 thing that does limit a 4500 is that there is "only" 6 Gbps of
    bandwidth between each blade and the Sup - might be an issue depending
    on your traffic patterns.
    >
    >Thanks in advance for your feedback.

    --
    Regards

    - replace xyz with ntl
     
    Stephen, Aug 23, 2008
    #2
    1. Advertising

  3. PurpleServerMonkey

    Merv Guest

    also keep in mind software failures ...

    If you have or can get the budget, then 2 switches is the way to go to
    improve the redundancy of your network along with dual homing of
    course
     
    Merv, Aug 23, 2008
    #3
  4. On Aug 24, 4:15 am, Stephen <> wrote:
    > On Sat, 23 Aug 2008 05:31:50 -0700 (PDT), PurpleServerMonkey
    >
    > <> wrote:
    > >We currently have a single Cisco Catalyst 4510 with dual supervior IV
    > >modules and dual power supplied. It's central to our network with
    > >workgroup switches and servers connecting to it.

    >
    > >My question is how internally redundant is a Cisco 4510?

    >
    > there are backplane traces to both Sups from each card, so you can
    > tolerate loss of most things as long as your network design does not
    > depend on a single I/O blade.
    >
    > >Basically investigating our options in making the core more fault
    > >tolerant but haven't found a lot of information around what could go
    > >wrong with a 4510. Trying to decide whether we should add another 4500
    > >series switch to the core or upgrade to something like a 6500.

    >
    > The old joke is that it is not resilient until it could survive a
    > nutter with a sledgehammer :) - but not many sites meet that spec...
    >
    > 6500 gets you more of everything - bandwidth and flexibility mainly
    > but you also get Netflow on some setups which might be importantm and
    > better inservice upgrade - but 1 box has inherent limitations.
    >
    > so i would go for 2 4510s as dual star points for the network - even
    > if they do not have dual sups.
    >
    > 1 thing that does limit a 4500 is that there is "only"  6 Gbps of
    > bandwidth between each blade and the Sup - might be an issue depending
    > on your traffic patterns.
    >
    > >Thanks in advance for your feedback.

    >
    > --
    > Regards
    >
    > - replace xyz with ntl


    Thanks for the info.

    We have deployed everything across multiple line cards so we have been
    rather happy with the single switch however we are going through the
    process of eliminating single points of failure.

    Budget isn't a problem so we might look at a 6500 series device so
    that we can go to 10GE modules.
     
    PurpleServerMonkey, Aug 24, 2008
    #4
  5. PurpleServerMonkey

    Scott Perry Guest

    Check your supervisor module failover configuration. Just because there are
    two supervisor cards does not mean that the best failover method possible
    happen. One failover method waits for the second supervisor module to boot
    while stateful switchover (SSO) quickly moves to the new supervisor card.

    Another failure when only the one modular switch is used, other than the
    software crashing, is a backplane failure. The idea of having 2 switches is
    good, especially when an etherchannel can interconnect them, except for all
    of the hosts which formerly connected to a single switch. That might be a
    problem to move in the event of a failure. One solution is to have single
    homed hosts connect to a third switch with dual uplinks into both 4500s.

    -----
    Scott Perry
    Indianapolis, IN
    -----

    "PurpleServerMonkey" <> wrote in message
    news:...
    On Aug 24, 4:15 am, Stephen <> wrote:
    > On Sat, 23 Aug 2008 05:31:50 -0700 (PDT), PurpleServerMonkey
    >
    > <> wrote:
    > >We currently have a single Cisco Catalyst 4510 with dual supervior IV
    > >modules and dual power supplied. It's central to our network with
    > >workgroup switches and servers connecting to it.

    >
    > >My question is how internally redundant is a Cisco 4510?

    >
    > there are backplane traces to both Sups from each card, so you can
    > tolerate loss of most things as long as your network design does not
    > depend on a single I/O blade.
    >
    > >Basically investigating our options in making the core more fault
    > >tolerant but haven't found a lot of information around what could go
    > >wrong with a 4510. Trying to decide whether we should add another 4500
    > >series switch to the core or upgrade to something like a 6500.

    >
    > The old joke is that it is not resilient until it could survive a
    > nutter with a sledgehammer :) - but not many sites meet that spec...
    >
    > 6500 gets you more of everything - bandwidth and flexibility mainly
    > but you also get Netflow on some setups which might be importantm and
    > better inservice upgrade - but 1 box has inherent limitations.
    >
    > so i would go for 2 4510s as dual star points for the network - even
    > if they do not have dual sups.
    >
    > 1 thing that does limit a 4500 is that there is "only" 6 Gbps of
    > bandwidth between each blade and the Sup - might be an issue depending
    > on your traffic patterns.
    >
    > >Thanks in advance for your feedback.

    >
    > --
    > Regards
    >
    > - replace xyz with ntl


    Thanks for the info.

    We have deployed everything across multiple line cards so we have been
    rather happy with the single switch however we are going through the
    process of eliminating single points of failure.

    Budget isn't a problem so we might look at a 6500 series device so
    that we can go to 10GE modules.
     
    Scott Perry, Aug 25, 2008
    #5
  6. PurpleServerMonkey

    Scott Perry Guest

    This is a quick failover configuration for the 4500:

    redundancy
    mode sso
    main-cpu
    auto-sync running-config
    auto-sync standard

    -----
    Scott Perry
    Indianapolis, IN
    -----

    "Scott Perry" <scott.perry@somecompany> wrote in message
    news:48b2b212$0$3711$...
    > Check your supervisor module failover configuration. Just because there
    > are two supervisor cards does not mean that the best failover method
    > possible happen. One failover method waits for the second supervisor
    > module to boot while stateful switchover (SSO) quickly moves to the new
    > supervisor card.
    >
    > Another failure when only the one modular switch is used, other than the
    > software crashing, is a backplane failure. The idea of having 2 switches
    > is good, especially when an etherchannel can interconnect them, except for
    > all of the hosts which formerly connected to a single switch. That might
    > be a problem to move in the event of a failure. One solution is to have
    > single homed hosts connect to a third switch with dual uplinks into both
    > 4500s.
    >
    > -----
    > Scott Perry
    > Indianapolis, IN
    > -----
    >
    > "PurpleServerMonkey" <> wrote in message
    > news:...
    > On Aug 24, 4:15 am, Stephen <> wrote:
    >> On Sat, 23 Aug 2008 05:31:50 -0700 (PDT), PurpleServerMonkey
    >>
    >> <> wrote:
    >> >We currently have a single Cisco Catalyst 4510 with dual supervior IV
    >> >modules and dual power supplied. It's central to our network with
    >> >workgroup switches and servers connecting to it.

    >>
    >> >My question is how internally redundant is a Cisco 4510?

    >>
    >> there are backplane traces to both Sups from each card, so you can
    >> tolerate loss of most things as long as your network design does not
    >> depend on a single I/O blade.
    >>
    >> >Basically investigating our options in making the core more fault
    >> >tolerant but haven't found a lot of information around what could go
    >> >wrong with a 4510. Trying to decide whether we should add another 4500
    >> >series switch to the core or upgrade to something like a 6500.

    >>
    >> The old joke is that it is not resilient until it could survive a
    >> nutter with a sledgehammer :) - but not many sites meet that spec...
    >>
    >> 6500 gets you more of everything - bandwidth and flexibility mainly
    >> but you also get Netflow on some setups which might be importantm and
    >> better inservice upgrade - but 1 box has inherent limitations.
    >>
    >> so i would go for 2 4510s as dual star points for the network - even
    >> if they do not have dual sups.
    >>
    >> 1 thing that does limit a 4500 is that there is "only" 6 Gbps of
    >> bandwidth between each blade and the Sup - might be an issue depending
    >> on your traffic patterns.
    >>
    >> >Thanks in advance for your feedback.

    >>
    >> --
    >> Regards
    >>
    >> - replace xyz with ntl

    >
    > Thanks for the info.
    >
    > We have deployed everything across multiple line cards so we have been
    > rather happy with the single switch however we are going through the
    > process of eliminating single points of failure.
    >
    > Budget isn't a problem so we might look at a 6500 series device so
    > that we can go to 10GE modules.
    >
    >
     
    Scott Perry, Aug 25, 2008
    #6
  7. Thanks for the info on the quick supervisor fail over, I double
    checked our configuration and it's set correctly.

    Backplane failure is probably our biggest concern, software crashing
    and having to reboot it is a problem but relatively quick to resolve
    whereas a backplane problem, although rare it has the potential to
    cause a major outage in our environment.

    We actually have the same configuration over a number of sites and are
    looking into the 6500 as a possible replacement versus adding another
    4500 to the mix.
     
    PurpleServerMonkey, Aug 25, 2008
    #7
    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. John Ireland
    Replies:
    2
    Views:
    609
    Javier
    May 10, 2004
  2. Replies:
    4
    Views:
    849
  3. ttripp
    Replies:
    5
    Views:
    4,671
    ttripp
    Jan 16, 2009
  4. ttripp
    Replies:
    1
    Views:
    899
    Doug McIntyre
    Dec 8, 2009
  5. spinnekop
    Replies:
    0
    Views:
    1,692
    spinnekop
    Jan 19, 2012
Loading...

Share This Page