Mailing List Archive: Cisco: NSP

An observation: 512k default max-prefix in IOS-XR

Index | Next | Previous | View Threaded

will at harg

Apr 25, 2012, 6:17 PM

Post #1 of 6 (975 views)
Permalink

An observation: 512k default max-prefix in IOS-XR

Configuring some ASR9ks, and I noticed:

http://www.cisco.com/en/US/docs/routers/asr9000/software/asr9k_r4.2/routing/configuration/guide/b_routing_cg42asr9k_chapter_00.html#con_1088375

""" To prevent a peer from flooding BGP with advertisements, a limit is placed on the number of prefixes that are accepted from a peer for each supported address family. The default limits can be overridden through configuration of the maximum-prefix limit command for the peer for the appropriate address family. The following default limits are used if the user does not configure the maximum number of prefixes for the address family:
• 512K (524,288) prefixes for IPv4 unicast """

524288 is also default setting for 6500/7600 PFC3BXL 'mls cef maximum routes'.

As usual when wondering about DFZ growth, one reaches for Geoff Huston's excellent presentations:
https://ripe64.ripe.net/presentations/25-2012-04-16-bgp2011.pdf

So around 2015 or perhaps before, careless operators of both IOS-XR and 7600 boxes may have a shock. :-)

You may wish to manually set max-prefix to something higher on your transit provider sessions. And, look at those -3BXL boxes too.

--
Will

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

rblayzor.bulk at inoc

Apr 28, 2012, 6:15 AM

Post #2 of 6 (905 views)
Permalink

Re: An observation: 512k default max-prefix in IOS-XR [In reply to]

On Apr 25, 2012, at 9:17 PM, Will Hargrave wrote:
> So around 2015 or perhaps before, careless operators of both IOS-XR and 7600 boxes may have a shock. :-)

Careless operators of 9K that don't change from the default scale maybe. That's just a default limitation on the 9K.

Just change from the "default" scale to l3 or l3xl scale and you'll be good for 1M+ routes. Of course L3 scale eats up L2 scale and vice versa, so choose your battles wisely.

--
Robert Blayzor
INOC, LLC
rblayzor [at] inoc
http://www.inoc.net/~rblayzor/

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

sigurbjornl at vodafone

Apr 28, 2012, 12:46 PM

Post #3 of 6 (898 views)
Permalink

Re: An observation: 512k default max-prefix in IOS-XR [In reply to]

8 years ago when Cisco introduced the 3BXL, 1M routes was a lot of routes,
more than 5 times the size of the current Internet routing table, now,
it's not.

1M routes also doesn't mean 1M routes. IPv6, Multicast and IPv6 Multicast
take up more space, further reducing the available tcam resources.

With an expanding IPv6 table and a never-ending fragmentation of the IPv4
space, plus whatever multicast and MPLS you are doing, it might prove to
be insufficient moving forward, and in some cases depending on what you're
doing, it might already be.

The L3XL kills off video monitoring and requires an even more expensive
license and this is one of the main reasons the ASR9k doesn't sound that
exciting to me at this point. I'd like to see a lot more available tcam
space before committing to a new platform long term.

Kind regards,
Sibbi

On 28.4.2012 13:15, "Robert Blayzor" <rblayzor.bulk [at] inoc> wrote:

>On Apr 25, 2012, at 9:17 PM, Will Hargrave wrote:
>> So around 2015 or perhaps before, careless operators of both IOS-XR and
>>7600 boxes may have a shock. :-)
>
>
>Careless operators of 9K that don't change from the default scale maybe.
>That's just a default limitation on the 9K.
>
>Just change from the "default" scale to l3 or l3xl scale and you'll be
>good for 1M+ routes. Of course L3 scale eats up L2 scale and vice versa,
>so choose your battles wisely.
>
>--
>Robert Blayzor
>INOC, LLC
>rblayzor [at] inoc
>http://www.inoc.net/~rblayzor/
>
>
>
>
>_______________________________________________
>cisco-nsp mailing list cisco-nsp [at] puck
>https://puck.nether.net/mailman/listinfo/cisco-nsp
>archive at http://puck.nether.net/pipermail/cisco-nsp/

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

rblayzor.bulk at inoc

Apr 28, 2012, 1:10 PM

Post #4 of 6 (900 views)
Permalink

Re: An observation: 512k default max-prefix in IOS-XR [In reply to]

On Apr 28, 2012, at 3:46 PM, Sigurbjörn Birkir Lárusson wrote:
> The L3XL kills off video monitoring and requires an even more expensive
> license and this is one of the main reasons the ASR9k doesn't sound that
> exciting to me at this point. I'd like to see a lot more available tcam
> space before committing to a new platform long term.

Well the reality of the 720-3BXL in an IPv4/v6 world is that you get about 500k IPv4 and half that in IPv6 (IIRC).

The l3xl scale license will get you 1M IPv4 routes and 128K IPv6. Of course you'll lose a lot of your L2 scale. So if your considering the 9K, it would make a decent Internet facing router or massive Ethernet transport/aggregation platform... not both. Maybe this changed with the new RSP's...

--
Robert Blayzor
INOC, LLC
rblayzor [at] inoc
http://www.inoc.net/~rblayzor/

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

jeff-kell at utc

Apr 28, 2012, 1:27 PM

Post #5 of 6 (902 views)
Permalink

Re: An observation: 512k default max-prefix in IOS-XR [In reply to]

On 4/28/2012 4:10 PM, Robert Blayzor wrote:
> Well the reality of the 720-3BXL in an IPv4/v6 world is that you get
> about 500k IPv4 and half that in IPv6 (IIRC). The l3xl scale license
> will get you 1M IPv4 routes and 128K IPv6. Of course you'll lose a lot
> of your L2 scale. So if your considering the 9K, it would make a
> decent Internet facing router or massive Ethernet
> transport/aggregation platform... not both. Maybe this changed with
> the new RSP's...

For a VS720/PFC3CXL default allocations:

UTC-Border#show platform hardware capacity forward

L3 Forwarding Resources
FIB TCAM usage: Total Used
%Used
72 bits (IPv4, MPLS, EoM) 524288
408427 78%
144 bits (IP mcast, IPv6) 262144
8 1%

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

philxor at gmail

Apr 28, 2012, 2:25 PM

Post #6 of 6 (951 views)
Permalink

Re: An observation: 512k default max-prefix in IOS-XR [In reply to]

If you are willing to spend the money on the latest gen cards, the Typhoon
linecards support 4M IPv4/2M IPv6 in the FIB along with 2M MACs. There
are no different scale profiles for the cards to monkey around with
either.

Not really part of this conversation but I just ran across this somewhat
amusing message on a Typhoon card.

LC/0/0/CPU0:Apr 29 22:17:59.355 : rsi_agent[306]:
%OS-RSI_AGENT-6-CARD_ROLE_CHANGE : Based on the card configuration/type,
the AFI IPv6 role of the card has changed from Invalid to Not Interested

Phil

On 4/28/12 3:46 PM, "Sigurbjörn Birkir Lárusson" <sigurbjornl [at] vodafone>
wrote:

>8 years ago when Cisco introduced the 3BXL, 1M routes was a lot of routes,
>more than 5 times the size of the current Internet routing table, now,
>it's not.
>
>
>1M routes also doesn't mean 1M routes. IPv6, Multicast and IPv6 Multicast
>take up more space, further reducing the available tcam resources.
>
>
>With an expanding IPv6 table and a never-ending fragmentation of the IPv4
>space, plus whatever multicast and MPLS you are doing, it might prove to
>be insufficient moving forward, and in some cases depending on what you're
>doing, it might already be.
>
>The L3XL kills off video monitoring and requires an even more expensive
>license and this is one of the main reasons the ASR9k doesn't sound that
>exciting to me at this point. I'd like to see a lot more available tcam
>space before committing to a new platform long term.
>
>Kind regards,
>Sibbi
>
>On 28.4.2012 13:15, "Robert Blayzor" <rblayzor.bulk [at] inoc> wrote:
>
>>On Apr 25, 2012, at 9:17 PM, Will Hargrave wrote:
>>> So around 2015 or perhaps before, careless operators of both IOS-XR and
>>>7600 boxes may have a shock. :-)
>>
>>
>>Careless operators of 9K that don't change from the default scale maybe.
>>That's just a default limitation on the 9K.
>>
>>Just change from the "default" scale to l3 or l3xl scale and you'll be
>>good for 1M+ routes. Of course L3 scale eats up L2 scale and vice versa,
>>so choose your battles wisely.
>>
>>--
>>Robert Blayzor
>>INOC, LLC
>>rblayzor [at] inoc
>>http://www.inoc.net/~rblayzor/
>>
>>
>>
>>
>>_______________________________________________
>>cisco-nsp mailing list cisco-nsp [at] puck
>>https://puck.nether.net/mailman/listinfo/cisco-nsp
>>archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>
>_______________________________________________
>cisco-nsp mailing list cisco-nsp [at] puck
>https://puck.nether.net/mailman/listinfo/cisco-nsp
>archive at http://puck.nether.net/pipermail/cisco-nsp/

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads