[Outages-discussion] [outages] Intermittent Level 3 outage/connectivity issues on Dallas edges?

Jeremy Chadwick jdc at koitsu.org
Thu Jan 24 18:02:55 EST 2013


On Thu, Jan 24, 2013 at 03:31:44PM -0500, Jay Ashworth wrote:
> [ moving to -discuss ]
> 
> ----- Original Message -----
> > From: "Jeremy Chadwick" <jdc at koitsu.org>
> 
> > What you showed in Item #3 (your TW Telecom link) **does** show packet
> > loss, but I cannot tell (without Item #4) if it's on the return-path or
> > the forward-path. I understand "your TW Telecom link is working fine"
> > but the 0.1-0.2% packet loss indicates it may not be as good as you
> > think. That may be a separate issue however, or may be by design
> > (QoS-like rules applied on router, etc.).
> > 
> > Make sense? :-)
> 
> It might make sense to him, or even to you, but it doesn't to me.
> 
> At each hop in an outbound trace, is it not true that the route which
> the reply packets will take back to you might itself be different, depending
> on *that device's* view of the routing tables, just as a trace from the 
> actual destination device back to the origination point may take a different
> route from the trace to it?
> 
> I never thought about that before you pressed the point just now, but ISTM
> that given asymmetrical routing, there's no guarantee that the pongs on a
> trace *to* something will come back along the path over which they got to
> those intermediate routers...

Jay, you're absolutely correct on this point.  I didn't organise my
Email decently, nor did I make my point in a coherent or correct manner.
Normally I'm better about this, as I'm sure you know by now.

The simple version of what I was trying to say, but failed, is: when
troubleshooting network issues, always assume routing is asymmetrical.
A single traceroute (A->B) can show "that something is happening" but
cannot accurately portray where, network-wise, the issue may** lie.
It can only give you a general idea of which providers/routers are
involved.

**: I say "may" because even with traceroutes in both directions you're
still blind to issues at higher layers (1 and 2).

-- 
| Jeremy Chadwick                                   jdc at koitsu.org |
| UNIX Systems Administrator                http://jdc.koitsu.org/ |
| Mountain View, CA, US                                            |
| Making life hard for others since 1977.             PGP 4BD6C0CB |


More information about the Outages-discussion mailing list