[PLUG] cisco 678 DSL latency issues

Derek Loree drl at drloree.com
Tue Aug 5 13:12:02 UTC 2003


On Tue, 2003-08-05 at 11:36, Rogan Creswick wrote:
> On Tue, Aug 05, 2003 at 11:19:44AM -0700, Derek Loree wrote:
> > I have a Cisco 678 running in bridged mode, and I do not have a 2 second
> > latency.  Have you installed the latest firmware on the 678?  
> 
> Yes, I installed 2.4.6 last night in hopes of fixing this
> problem. Previously it was running 2.4.1 (CBOS).
> 
> > Are you
> > giving the connection enough time to straighten out the ARP tables
> > (anywhere from 5 to 15 minutes)?  
> 
> The problem has persisted for hours of "use" with enough idle time
> that I don't think this is a problem, but I have probably not turned
> on a router and let it sit for 15 min wlout any network activity.
> 
> > Have you tried to power down, wait 2
> > minutes, power up (this should reset the connection at the other end)? 
> 
> This has no effect.
> 
> > BTW, download rates should peak at about 68k/s.
> 
> nods, once established, file transfers are at expected speeds. 
> 
> 
> > > 
> > > I've tried three computers and two phone jacks, two of the computers
> > > dual boot (the third is my router, and runs only linux--everything is
> > > Debian).  In all situations Windows works fine (90 ms ping) and Linux
> > > does awefully (2000ms).  
> > 
> > Are you saying that each ping packet takes 2 sec. to return, or just the
> > first packet?  Does tcptraceroute show where the slow down is? 
> 
> Each packet takes 2 sec. to return. The delay is between my cisco 678
> and the qwest gateway.  I can't go _anywhere_ without this delay
> (unless I'm in windows, then it all works fine).

How does Qwest fit into this?  I thought your ISP was Unicom, that is
where your gateway should be.  Qwest should only be providing the low
level transport to your ISP.
> 
> This is also not an icmp only problem, tcp is also affected (I dont
> use ipx).
> 
> > > effect (possibly a _small_ change, 1800ms is probably the average).
> 
> This change was a freak of nature, still at 2000ms. (Just a FYI).
> 
> > This seems like more of an ISP problem than a Cisco problem.  In
> > bridging mode, the Cisco should let anything and everything pass right
> > through.
> 
> I would agree entirely, if Windows didn't work flawelessly in this regard.

Which makes me think that something is not configured correctly on your
linux boxes, but I can't imagine what it would be.

Sorry I can't be more helpful.

Derek Loree





More information about the PLUG mailing list