Re: [Wireshark-users] Question on InternetPerformanceTroubleshooting
But MLPPP can definitely be a source of out of order delivery! Mike On Mar 12, 2007, at 5:29 PM, Stephen Fisher wrote: > On Fri, Mar 02, 2007 at 05:01:09PM -0500, Small, James wrote: > >> I believe the 3 T1 are multiplexed using multilink PPP using an >> Adtran >> router if I remember correctly. >> >> Is there any way to tell if this PPP bundle is causing out of order >> packets or other issues? > > Not really that I know of. :( ___ Wireshark-users mailing list Wireshark-users@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-users
Re: [Wireshark-users] Question on InternetPerformanceTroubleshooting
On Fri, Mar 02, 2007 at 05:01:09PM -0500, Small, James wrote: > I believe the 3 T1 are multiplexed using multilink PPP using an Adtran > router if I remember correctly. > > Is there any way to tell if this PPP bundle is causing out of order > packets or other issues? Not really that I know of. :( Steve ___ Wireshark-users mailing list Wireshark-users@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-users
Re: [Wireshark-users] Question on InternetPerformanceTroubleshooting
Hi Sake, Not an unreasonable suspicion - in fact, when I used: http://miranda.ctd.anl.gov:7123/ The site suspected a duplex mismatch since my download speed tends to be less than half of my upload speed. Many times the upload speed is close to the advertised rate but I have never been able to get the full download speed. Maybe I can double check with the provider on their router - but they said they already checked everything and the service provider seems decent. Still, it's probably worth double checking. On all my equipment, there are no errors/FCS, drops, out of buffers - everything is perfect (from an Ethernet stand point anyway). The newer stuff is gigabit where the IEEE mandates auto-negotiation in the spec. The older stuff that's 100 Mbps is hard coded just like you said. I guess if it were easy there wouldn't be a whole IT profession, eh? :-) --Jim > You probably have checked this already, but I could not resist in > mentioning it, did you check the duplex settings on the uplink-router, > the firewall and the switch-ports? If the packet-loss is higher when > your (local) traffic increases, but your traffic is not maxing out > your links, it does sound like a local problem and duplex mismatches > are still source nr.1 in my experience. > > If it is possible, set all speeds and duplex-modes fixed. Having one > side on fixed and the other side on auto is a sure cause for trouble. > Having both sides on auto usually works, but does indeed give you > duplex-mismatches sometimes. If you have a duplex mismatch, you will > see a lot of FCS/alignment errors on the interface in full-duplex > mode and a lot of collisions on the interface in half-duplex mode. ___ Wireshark-users mailing list Wireshark-users@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-users
Re: [Wireshark-users] Question on InternetPerformanceTroubleshooting
Steve, I believe the 3 T1 are multiplexed using multilink PPP using an Adtran router if I remember correctly. Is there any way to tell if this PPP bundle is causing out of order packets or other issues? Thanks, --Jim > > One off the wall idea - the site had two T1's (3.0 Mbps) multiplexed > > via PPP before. The problems seem to start close to around when they > > added a third T1 (again via PPP) for a total of approx 4.5Mbps. Is > > there any chance that this could cause issues - seems to be a pretty > > standard provider setup... > > How are the three T1s load-balanced? Multilink PPP or just using three > paths that the routers see between each other? When there are just > three paths seen between the routers, the routers will often cache which > destination goes over which circuit so the packets are transmitted > across the same circuit in proper order for each destination on the > other end. Multilink PPP sends the packets in more of a round-robin > fashion, where one of the packets could get caught behind a larger > packet on say the first T1 while two other packets from the same session > make it across the other two T1s quickly. This would cause out-of-order > packets. Although that case is usually confined to slower speed links > (< 768Kbps) and is called serialization delay. ___ Wireshark-users mailing list Wireshark-users@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-users
