Re: [mpls-linux-general] delay Difference at Ingress and Egress

[James R. L. <jl...@mi...>]

I would expect Ingress to be slower, I just didn't think it would
be that slow :-)

Another thing to note is that when using tcpdump or the like, your
forcing the MPLS stack to make a private copy (otherwise the version
seen by tcpdump will be corrupt).  So you are forcing some aditional
mem copies.

Armed with this new info I'll see if I can find some way to optimize
the ingress LER case.

Thanks for your work.

On Sun, Dec 08, 2002 at 03:27:25PM -0800, nitin panjwani wrote:
> Hi James,
> 
> Recently I measured the processing delay caused
> protocol stack at Ingress, Egress and core.
> 
> Here are my results:
> Ingress: 3.665 ms
> Egress: 0.035 ms
> Core: 0.018 ms
> IP:0.023 ms
> 
> It makes sense that processing time at these nodes
> will be in the following order
> 
> Ingress >= Egress > Core
> 
> What I am wondering, is huge delay at Ingress. If we
> do L3 lookup + L2.5 forwarding even then it doesn't
> clearify the reason of huge value. Can you please
> explain in short how processing is taking place at
> Ingress and how it can explain this delay.
> 
> Let me also explain my measuring technique, you can
> corect me on this if I am doing something wrong.
> 
>       ---e1[Node]e0------
> 
> In this I am running tcpdump/tethereal at both e0 and
> e1. What I have understood is that libcap(used by
> tcpdump/tethereal) captures the packet at device
> driver, which is after it has been processed by
> protocol stack for outgoing packet and before being
> processed by protocol stack for incomming packet.  
> 
> Thanks and Regards,
> Nitin
> 
> 	
> 					
> 
> 
>     
> 
> 
> 
> 
> __________________________________________________
> Do you Yahoo!?
> Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
> http://mailplus.yahoo.com
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> https://lists.sourceforge.net/lists/listinfo/mpls-linux-general

-- 
James R. Leu