[nsp] 7500 insertion force

Charles Sprickman spork@inch.com
Sun, 8 Sep 2002 02:32:15 -0400 (EDT)


On Sat, 7 Sep 2002, Rafi Sadowsky wrote:

>  What about 720X OIR experiences ?

Has been extremely kind to me.  Both in production gear and in a few
offline tests where I've rearranged 5 cards to see what would happen.

Charles

>
>  Please see in-line comments below for more detail
>
> --
> Thanks
> =09Rafi
>
>
>
> Date: Fri, 6 Sep 2002 15:27:31 -0600
> From: Clinton Work <work@scripty.com>
> To: Cisco-NSP <cisco-nsp@puck.nether.net>
> Subject: Re: [nsp] 7500 insertion force
>
>
> The Cisco 7500 CBUS is shared amoung all the line cards attached
> to it. When you do an OIR the bus is electrically interrupted and a
> CBUS complex operation is started to rediscover the cards. This usually
> means that all packet forwarding stops for 20-30 seconds. If your lucky,
> everything comes back and the new card is ready for use. We have run
> into a lot of problems with OIR and we now power off our 7513s before swa=
pping
> line cards.
>
> CBUS complex:
> http://www.cisco.com/warp/public/63/cbus_complex.html
> http://www.cisco.com/warp/public/63/top_issues/router_issues/cbus_complex=
=2Ehtml
>
>  Quoting from the end of the second URL refernce:
>
> ---
> How Can I Avoid a Cbus Complex When a VIP Reloads?
>
>  A new feature called Single Line Card Reload (SLCR) has recently been
> introduced in Cisco IOS=AE Software version 12.0(13)S. This feature allow=
s
> you to only reset the affected line card, as opposed to resetting all the
> line cards in the system, which results in a shorter recovery time, and i=
n
> a much shorter traffic interruption.
>
>  When this feature is enabled:
>
>   * The affected card first downloads its microcode without blocking the
>  traffic between other interface processors.
>
>   * The MEMD is then re-carved. As stated above, this process only takes
>  a few seconds. During that time, all traffic is interrupted.
>
>  In summary, traffic interruption lasts around 20 seconds (microcode
> download and re-carving) for the affected interface, and just a few
> seconds for all other interfaces (compared to two minutes for all the
> interfaces in a standard Cbus Complex).
> ---
>
>  Quoting from:
>
> <http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/120newft=
/120limit/120s/120s13/slcr.htm>
> ---
>  Feature Overview
>
>  Before the introduction of the Cisco 7500 Single Line Card Reload
> feature, the only method of correcting a line card hardware failure or a =
severe
> software error for one line card on a Cisco 7500 series router required
> the execution of a Cbus Complex, a process that reloaded every line card
> on the network backplane. The time taken to complete the Cbus Complex was
> often inconvenient, and no network traffic could be routed or switched
> during the Cbus Complex process.
>
>  The Cisco 7500 Single Line Card Reload feature allows users to correct a
> line card failure on a Cisco 7500 series router by automatically reloadin=
g
> the microcode on a failed line card. During the single line card reload
> process, all physical lines and routing protocols on the other line cards
> of the network backplane remain active. A single line card reload is also
> significantly faster than the Cbus Complex process.
> Benefits
>
> Improved Line Card Recovery Time
>
>  The time required to correct a line card hardware failure for one line
> card is significantly improved when the Cisco 7500 Single Line Card Reloa=
d
> feature is enabled. The entire system, which now only reloads one line
> card instead of every line card, also experiences a dramatic improvement
> in recovery time.
>
> Network Traffic Flow Improvements
>
>  Because the Cisco 7500 Single Line Card Reload feature only reloads the
> line card with the hardware failure rather than all of the line cards on
> the Cisco 7500 network backplane, the active line cards can continue to
> forward network traffic.
> ---
>
>
> _______________________________________________
> cisco-nsp mailing list  real_name)s@puck.nether.net
> http://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>