[Xorp-users] Multiple xorp instances performance improvements.

[Kristian Larsson]

Ben Greear wrote:
> Pavlin Radoslavov wrote:
>> Ben Greear <greearb at candelatech.com> wrote:
>>
>>   
>>> It appears that each Xorp instance effectively listens for all xorp
>>> related packets on all interfaces.  It does filter out the packets that
>>> are received on interfaces it is not configured for, but that is a lot
>>> of work.
>>>     
>> I presume you are talking about the raw IP interface.
>>   
> I think so..I haven't followed the code far enough to see exactly how
> it is created/bound yet.
>>> If we are talking about 10 or 100 xorp instances, then each router packet
>>> is going to wake 10 or 100 processes (9 or 99 of which don't care),
>>> causing serious performance penalty.
>>>     
>> This might or might not be the case, but I wouldn't call that
>> "serious performance penalty" without doing some profiling or
>> measurements that quantify it.
>>   
> Fair enough, though if we reach 50 xorps, with each sending 1 pkt per 
> second, that means
> that all 50 will be receiving 50 per second..and this is about what VOIP 
> costs.  I do know that
> running 50 VOIP calls on a system will drag down a single CPU easily.

I'm not really in to this discussion, but...
50 VOIP calls does not equal 50 packets per second. Let's say we are 
using G.729 for coding our voice, it's a 8Kbps codec and we use cRTP 
(compressed RTP). G.729 sample size being 10ms this equals 80 bits per 
sample. It's quite normal to use two samples per packet, ie 20 bytes 
which in turn would equal somewhere around 50 packets per second. Now 50 
packets per second is no problem for a normal computer. Everyone is 
running Skype or something like it and the processing of that is no 
problem.
Not even 50 calls, ie 2500 packets per second should be any trouble to 
handle as long as you're not doing any transcoding... it's the coding 
stuff that's CPU intense.
Just for a reference note, I've ran computers as media gateways with 120 
  simultaneous calls. They were loaded but ran fine.
Anyway, enough jibberish about VOIP. I don't think 50 packets per second 
should pose any problem at all to XORP and so I'm with Pavlin. We need 
some real measurement data to show that there is actually a problem.

Cheers,
    Kristian.