Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Quagga: Users

iBGP and eBGP peers

 

 

Quagga users RSS feed   Index | Next | Previous | View Threaded


sclark at netwolves

Mar 21, 2012, 8:15 AM

Post #1 of 18 (2049 views)
Permalink
iBGP and eBGP peers

Hello,

I received this question from a cisco engineer:
How many iBGP and eBGP peers does quagga support?

What would be the correct answer?


--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com


nick at inex

Mar 21, 2012, 8:37 AM

Post #2 of 18 (2015 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 21/03/2012 15:15, Steve Clark wrote:
> I received this question from a cisco engineer:
> How many iBGP and eBGP peers does quagga support?
>
> What would be the correct answer?

42.

Nick

_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


sclark at netwolves

Mar 21, 2012, 8:58 AM

Post #3 of 18 (2013 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 03/21/2012 11:37 AM, Nick Hilliard wrote:
> On 21/03/2012 15:15, Steve Clark wrote:
>> I received this question from a cisco engineer:
>> How many iBGP and eBGP peers does quagga support?
>>
>> What would be the correct answer?
> 42.
>
Darn I should have remembered that from the HHGTTG. ;-)

Is this really a resource issue and not a physical number thing?


--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com


gdt at ir

Mar 21, 2012, 9:15 AM

Post #4 of 18 (2012 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

Is this really a resource issue and not a physical number thing?

Do you realize that quagga is just software? So the answer is really
that it depends on a vast number of things, including which OS you use
and which hardware, the size of tables you get from each peer, etc.
Given that, your question asking for a simple answer doesn't make any
sense.


sclark at netwolves

Mar 21, 2012, 9:25 AM

Post #5 of 18 (2008 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 03/21/2012 12:15 PM, Greg Troxel wrote:
> Is this really a resource issue and not a physical number thing?
>
> Do you realize that quagga is just software? So the answer is really
> that it depends on a vast number of things, including which OS you use
> and which hardware, the size of tables you get from each peer, etc.
> Given that, your question asking for a simple answer doesn't make any
> sense.
>
Hi Greg,

Thanks for the response. I was asked the question by our "new to the company" network architect.

We have an inhouse built VPN/FW/ROUTER built appliance that runs on a 1.66ghz dual core
dual thread atom processor with 2gb of memory, with centos 6.2 and quagga 0.99.18.

He is not enamored with our in house platform since it isn't a cisco.

So I am just trying to get some real world information to provide him.


--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com
_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


gdt at ir

Mar 21, 2012, 9:30 AM

Post #6 of 18 (2010 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

I would suggest that you test your box, rather than asking here.....


sclark at netwolves

Mar 21, 2012, 9:36 AM

Post #7 of 18 (2010 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 03/21/2012 12:30 PM, Greg Troxel wrote:
> I would suggest that you test your box, rather than asking here.....
>
Hi Greg,

Is there a limit like 1, 2, 4 etc or not simple question?

--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com


gdt at ir

Mar 21, 2012, 9:38 AM

Post #8 of 18 (2015 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

Well, you asked "how many". I would be extremely surprised if you had
trouble with 4, or even 8. But if you ask about 64 full BGP feeds, I
don't know.


sclark at netwolves

Mar 21, 2012, 10:05 AM

Post #9 of 18 (2011 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 03/21/2012 12:38 PM, Greg Troxel wrote:
> Well, you asked "how many". I would be extremely surprised if you had
> trouble with 4, or even 8. But if you ask about 64 full BGP feeds, I
> don't know.
Hi Greg,

Thanks for taking the time to respond and have great day.

Regards,
--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com


tyler at tolaris

Mar 21, 2012, 10:05 AM

Post #10 of 18 (2015 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 2012-03-21 16:25, Steve Clark wrote:
> He is not enamored with our in house platform since it isn't a cisco.

Ah, the "Cisco guy" problem. I've got one of those.

I have had Quagga devices running on commodity PCs that handle full BGP
tables from multiple peers, with no problems at all. The only down side is
that has mattered to me has been (1) only one OSPF instance, and (2) the OS
needs more rebooting for applied security updates.

I doubt any amount of numbers or anecdotal evidence will sway him. But you
can override his opinion by showing the bosses the difference in cost.
Don't forget to include software updates and additional feature costs in
your Cisco pricing - both free with a Linux solution.

Regards,
Tyler

--
There is no "eastern" and "western" medicine. There's "medicine" and
then there's "stuff that has not been proven to work."
-- Maki Naro, "The Red Flags of Quackery, v2.0", Sci-ence.org
_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


stu at actusa

Mar 21, 2012, 10:13 AM

Post #11 of 18 (2017 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

I've been following this thread with great interest, as I am aware that
in most cases the limitation of [ie]BGP peers would have more to do with
ram and route count then anything else...

The question I'd like to ask is if there is a 'hard coded' maximum
number of BGP peers in the Quagga code itself. And while I'm asking, Is
there any hard coded maximum route count / buffer space in the code as
well...

Again, I pose the question from strictly a code standpoint... leaving
the OS and hardware abilities out, as those can be tested in the bench.

Thanks to all the developers for all the great work!

Stu

Stuart Sheldon
ACT USA


On 03/21/2012 09:38 AM, Greg Troxel wrote:
>
> Well, you asked "how many". I would be extremely surprised if you had
> trouble with 4, or even 8. But if you ask about 64 full BGP feeds, I
> don't know.
>
>
>
> _______________________________________________
> Quagga-users mailing list
> Quagga-users [at] lists
> http://lists.quagga.net/mailman/listinfo/quagga-users

- --
"Sometimes I lie awake at night and I ask, "Is life a multiple choice test
or is it a true or false test?" ...Then a voice comes to me out of the
dark and says, "We hate to tell you this but life is a thousand word essay."
-- Charles M. Schulz
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
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=q5l7
-----END PGP SIGNATURE-----
_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


tyler at tolaris

Mar 21, 2012, 10:21 AM

Post #12 of 18 (2011 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 2012-03-21 16:36, Steve Clark wrote:
> Is there a limit like 1, 2, 4 etc or not simple question?

I believe the limit is INT_MAX. RAM is the real limit.

Regards,
Tyler

--
"If we don't believe in free expression for people we despise, we don't
believe in it at all."
-- Noam Chomsky
_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


sclark at netwolves

Mar 21, 2012, 10:27 AM

Post #13 of 18 (2009 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 03/21/2012 01:21 PM, Tyler J. Wagner wrote:
> On 2012-03-21 16:36, Steve Clark wrote:
>> Is there a limit like 1, 2, 4 etc or not simple question?
> I believe the limit is INT_MAX. RAM is the real limit.
>
> Regards,
> Tyler
>
Hi Tyler,

Thanks for your input.

Regards,
--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com


scharf at isc

Mar 21, 2012, 3:03 PM

Post #14 of 18 (2009 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

Stephen,

OK, I'll actually try to help a bit. My testing was for default free BGP
feeds (~360k routes with "typical" update rates) with full cross
distribution and route injection into linux (note that all these numbers
are for the control plane processing, traffic forwarding was not tested
at the same time for this work.) We found we ran out of CPU on the
unthreaded bgpd long before we ran out of memory. With our given
snapshot, we were able to get 18 peers with full cross feed (adding peer
18 resulted in >2 minutes of cpu time for bgpd before returning to a
more normal state.) RAM use was still under 4GB, but a 3GHz xeon was out
of gas. Take that as one data point. I know that there are quagga based
route servers with >100 peers out there, so that's another data point.

Ask the Cisco advocate to price out a system that has enough ram and cpu
to handle 15 default free peers with cross feed.

I have not done the detailed research for how memory access played into
it, but it seems like a threaded bgpd could greatly improve the numbers.

jerry

On 03/21/2012 10:27 AM, Steve Clark wrote:
> On 03/21/2012 01:21 PM, Tyler J. Wagner wrote:
>> On 2012-03-21 16:36, Steve Clark wrote:
>>> Is there a limit like 1, 2, 4 etc or not simple question?
>> I believe the limit is INT_MAX. RAM is the real limit.
>>
>> Regards,
>> Tyler
>>
> Hi Tyler,
>
> Thanks for your input.
>
> Regards,
> --
> Stephen Clark
> *NetWolves*
> Director of Technology
> Phone: 813-579-3200
> Fax: 813-882-0209
> Email: steve.clark [at] netwolves
> http://www.netwolves.com
>
>
> _______________________________________________
> Quagga-users mailing list
> Quagga-users [at] lists
> http://lists.quagga.net/mailman/listinfo/quagga-users


nick at inex

Mar 22, 2012, 3:47 AM

Post #15 of 18 (2006 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

On 21/03/2012 17:13, Stuart Sheldon wrote:
> The question I'd like to ask is if there is a 'hard coded' maximum
> number of BGP peers in the Quagga code itself. And while I'm asking, Is
> there any hard coded maximum route count / buffer space in the code as
> well...

No - this is open source software so there are no artificial limitations
built in.

I haven't done a whole lot of ibgp testing, but from the route-server point
of view, my base test-lab configuration is ~250 clients with a smallish
number of prefixes each. This works fine with low churn levels.

There has been more extensive testing performed by some other people in the
european ixp scene. Elisa Jasinska wrote up a very brief summary of some
of the testing here:

http://ripe60.ripe.net/presentations/Jasinska-_Ab_Using_Route_Servers.pdf

The stability problems noted with quagga have been eradicated from Chris
Hall's euro-ix quagga branch. I did a bunch of testing over several days
last year and couldn't crash the daemon.

As a general comment, I will say that Cisco's BGP implementation scales
better than anyone else that I've seen and if you have the RAM for it, the
scability on the ASR1k platform is awesome. However, bgp stack scalability
isn't everything and if you want to get comparable generic router
performance to a quagga configuration like yours, you're going to end up
having to get a non-baseline asr1000 which will set you back considerably
more money than what you're currently running with.

If you're happy spending the money and understand what is different about
an asr1k over a server + quagga and still feel that an asr1k is a good use
of money (and in many cases, it is), then go for it. But using the number
of bgp peers supported as a metric of each platform's capability is about
as useful as comparing a toyota and a volkswagen on the basis of the size
of the windscreen washer fluid bottle.

Nick
_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


jorge at dti2

Mar 22, 2012, 4:35 AM

Post #16 of 18 (2008 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

El 22/03/2012 11:47, Nick Hilliard escribió:
> On 21/03/2012 17:13, Stuart Sheldon wrote:
>> The question I'd like to ask is if there is a 'hard coded' maximum
>> number of BGP peers in the Quagga code itself. And while I'm asking, Is
>> there any hard coded maximum route count / buffer space in the code as
>> well...
>
> No - this is open source software so there are no artificial limitations
> built in.
>
> I haven't done a whole lot of ibgp testing, but from the route-server point
> of view, my base test-lab configuration is ~250 clients with a smallish
> number of prefixes each. This works fine with low churn levels.
>
> There has been more extensive testing performed by some other people in the
> european ixp scene. Elisa Jasinska wrote up a very brief summary of some
> of the testing here:
>
> http://ripe60.ripe.net/presentations/Jasinska-_Ab_Using_Route_Servers.pdf
>
> The stability problems noted with quagga have been eradicated from Chris
> Hall's euro-ix quagga branch. I did a bunch of testing over several days
> last year and couldn't crash the daemon.
>

Hi, is this publicly available somewhere?

Regards,
Jorge

_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


chris.hall.list at highwayman

Mar 22, 2012, 5:42 AM

Post #17 of 18 (2010 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

Jorge Boncompte wrote (on Thu 22-Mar-2012 at 11:35 +0000):
> El 22/03/2012 11:47, Nick Hilliard escribió:
...
> > The stability problems noted with quagga have been eradicated from
> > Chris Hall's euro-ix quagga branch. I did a bunch of testing over
> > several days last year and couldn't crash the daemon.

> Hi, is this publicly available somewhere?

I am looking into a more public host, but if you wish to try out the
euro_ix version, please send me an ssh key and I will be happy to send
the required spells to reach the current git repository.

Chris


_______________________________________________
Quagga-users mailing list
Quagga-users [at] lists
http://lists.quagga.net/mailman/listinfo/quagga-users


sclark at netwolves

Mar 22, 2012, 7:34 AM

Post #18 of 18 (2005 views)
Permalink
Re: iBGP and eBGP peers [In reply to]

Hello,

Just wanted to thank everyone that took the time to respond to my query.

Regards,

--
Stephen Clark
*NetWolves*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark [at] netwolves
http://www.netwolves.com

Quagga users RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.