From: Basel Tashkandi (basel@xxxxxxxxxxxxx)
Date: Sat Dec 08 2001 - 20:29:28 GMT-3
Hi
I just want to add something that might help.
The delay is not necessary be due to the network. It can be an application
problem. It can be a Server or a client problem.
Try simulating the remote office by having another router connected
back-to-back with your main router or even a couple of new routers and see
if you still get the same poor performance even with higher speeds. This is
if the customer is welling to try that.
If the performance changes then check out for time-outs on the application
side it might be delay sensitive like SNA as you are aware.
Regards
At 15:08 08/12/2001 -0800, Yonkerbonk wrote:
> Well, I'm not saying that the broadcasts are the
>entire cause of the slow links, but I thought perhaps
>it was a contributing factor - especially the 56K
>links.
> Mainly, the client is running Citrix which needs 20K
>per connection. Some of the smaller sites with 56K I
>can see why they're slow, but some of the sites with
>384K and not that many users are also slow.
>I've tried extended pings with varying packet sizes
>and data patterns and they all come back relatively
>clean - 98% plus. Delay is not bad either. Only thing
>I can see is on some serial interfaces there are FECNs
>and the hub site shows BECNs, but it's only 2-3% of
>total inbound packets. But I don't see DE set, so
>nothing should be getting dropped. Some of the
>Ethernet intefaces also has high collisions, 15%, and
>I'll take care of that too.
> Anyways, the client was just wanting me to make a
>general diagnosis of his network. There are little
>things here and there that could contribute.
>I will put a WAN Sniffer on there and find out more.
> Thanks for the reply.
>
>Michael Le, CCIE #6811
>
>--- PimpDaddy <ccie@stephendunn.com> wrote:
> > 1.5MB*8/(24*3600)=138bits/sec.
> >
> > That sure aint a whole lot if you ask me. Probably
> > mostly comprised of
> > keepalives. Could be related to latency or dropped
> > packets, though. What
> > do your interface counters look like? What app
> > appears slow? What does a
> > continuous ping between the client station and app
> > server look like?
> >
> >
> > Steve
> >
> > > I have a client complaining of slow WAN links.
> > He
> > > has 3 hub routers, each with about 5-7 remote
> > links.
> > > They are all running OSPF. The remote sites are
> > not
> > > configured as stub areas even though there is
> > nothing
> > > behind them (I will remedy that).
> > > What I noticed was that the remote router Serial
> > > interfaces all had high 'out bcast bytes', that
> > > averaged out to 1.5 MB a day. Some of those
> > routers
> > > have only been up 5 days and some of them have
> > been up
> > > 15 weeks, but somehow they have all managed to
> > average
> > > 1.5 MB a day. And the only thing I can think of
> > that
> > > would show up as broadcasts on the Serial
> > interface is
> > > OSPF and CDP.
> > > They only have 2 subnets listed under OSPF with
> > a
> > > total of 50 routes when I do a show ip route. The
> > show
> > > ip ospf database is relatively small too - I only
> > need
> > > to hit spacebar once to see entire database.
> > > Anyone have any ideas if 1.5 MB is high? I think
> > it
> > > is since there are not alot of routes and there
> > > shouldn't be any changes since it's a stub router.
> > The
> > > 30 minute updates and CDP shouldn't contribute
> > much
> > > either.
> > > Thanks.
> > >
> > > Michael Le, CCIE #6811
> > >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:32:40 GMT-3