Cogent service

Matt Ryan Matt.Ryan at telewest.co.uk
Fri Sep 20 17:45:27 UTC 2002


I corrected myself to note this figure was us not ns (but I don't see that
post yet). When you say 1-2 ms added under load do you mean CPU load or
circuit load? Remember anytime the router has to buffer packets you are
going to see delay (unless you use a QoS mechanism - but that's a whole
other thread). Your example appears to backup my post - when you go a long
way around (despite the numbers of routers involved) it takes longer to get
there than the short route, but this isn't the routers making it slow, its
the distance.


Matt.

-----Original Message-----
From: David Diaz [mailto:davediaz at smoton.net]
Sent: 20 September 2002 18:35
To: Matt Ryan; William B. Norton; Ralph Doncaster; nanog at merit.edu
Cc: ml at vayner.net
Subject: RE: Cogent service


I dont believe that number at all.  Not to disagree with you at all 
Matt.  But real world, it would depend on the design.  If you are 
going router to router then it would be closer to accurate.  I have 
seen 1-2ms added per router under load.  That also might factor in a 
hop inside the room between 1 routers or another gear and a router. 
Coming in one card, having to do a lookup, going out another card to 
another box then out the layer 1 circuit to the next city.

A VPN might not save this time though.  But hubbing to a regional 
center, and then placing this traffic on an express circuit to the 
next regional city might save some of these ms.

In other words LA to San Fran, an express fiber route to Chi, then 
out a peering connection.  This would save 5-8ms compared to LA, San 
Fran, Portland, Salt Lake, Chi where at each hope 1-2 routers would 
insert themselves into the process.

In my book, 5-8ms is not a real factor....

Dont get me started on speed of light on fiber...

At 18:21 +0100 9/20/02, Matt Ryan wrote:
>>The hop count question is interesting.  Is the consensus that it's
>>mostly a customer service issue, where latency isnt affected but
>>customer perception is?  Or is it a real latency issue as more
>>routers take a few CPU cycles to make a routing decision.
>
>The routers these days make a forwarding decision in ~20 ns - its going to
>take a few hops before this becomes significant. Speed of light on long
>links is a couple of orders of magnitude more delay inducing.
>
>
>Matt.
>
>---------------------------------------------------------------------------
---
>Live Life in Broadband
>www.telewest.co.uk
>
>
>The information transmitted is intended only for the person or 
>entity to which it is addressed and may contain confidential and/or 
>privileged material.
>Statements and opinions expressed in this e-mail may not represent 
>those of the company. Any review, retransmission, dissemination or 
>other use of, or taking of any action in reliance upon, this 
>information by persons or entities other than the intended recipient 
>is prohibited. If you received this in error, please contact the 
>sender immediately and delete the material from any computer.
>
>
>===========================================================================
===

-- 

David Diaz
dave at smoton.net [Email]
pagedave at smoton.net [Pager]
Smotons (Smart Photons) trump dumb photons



------------------------------------------------------------------------------
Live Life in Broadband
www.telewest.co.uk


The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material.
Statements and opinions expressed in this e-mail may not represent those of the company. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender immediately and delete the material from any computer.


==============================================================================




More information about the NANOG mailing list