[122680] in North American Network Operators' Group
RE: MLFR Differential Delay Problems
daemon@ATHENA.MIT.EDU (Pender, James)
Fri Feb 19 11:47:39 2010
From: "Pender, James" <James.Pender@PAETEC.com>
To: "R. Benjamin Kessler" <rbk@mnsginc.com>, "nanog@nanog.org"
<nanog@nanog.org>
Date: Fri, 19 Feb 2010 11:47:02 -0500
In-Reply-To: <0B608FCDCF43BF4C9194C896C532024371A4D4@mnsg-svr2.mnsg.net>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
The differential delay is most likely caused by the T1's in the MFR bundle =
riding physically diverse paths across the TDM network. The carrier would n=
eed to validate their CLR/DLR to see what paths/DS3's the individual T1's f=
ollow to verify they are on the same circuit. Unfortunately there are those=
that try and sell MFR as "redundancy" and have the T1's ride diverse paths=
that can sometimes be pretty huge in difference of loop distance etc.., wh=
en they should really just be selling MFR for the bandwidth.=20
- Jim P.=20
-----Original Message-----
From: R. Benjamin Kessler [mailto:rbk@mnsginc.com]=20
Sent: Thursday, February 18, 2010 10:55 PM
To: nanog@nanog.org
Subject: MLFR Differential Delay Problems
Hello NANOGers -=20
=20
I'm working on a project to migrate a customer from one "Tier 1"
provider to another at 50+ locations (all domestic US sites). Most of
these connections are 4xT1 multi-link bundles.
=20
The old router configuration was MLPPP which was rock-solid for 3 years
(save for the typical "last-mile" circuit issues, fiber-cuts, etc.).
The new carrier uses FRF.16 multi-link Frame Relay vs. MLPPP.
=20
We've completed the migration on 10+ sites and all of them are now
reporting errors like the following:
=20
Feb 17 21:01:39 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/0
differential 91.7 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/0
differential 115.9 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/1
differential 79.0 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/1
differential 79.1 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/1
differential 97.4 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/0
differential 97.5 ms over yellow differential delay 75 ms
Feb 17 21:01:50 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/1
differential 97.5 ms over yellow differential delay 75 ms
Feb 17 21:01:52 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/1
differential 97.4 ms over yellow differential delay 75 ms
Feb 17 21:01:52 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/0
differential 97.5 ms over yellow differential delay 75 ms
Feb 17 21:01:52 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/1
differential 97.5 ms over yellow differential delay 75 ms
Feb 17 21:01:53 /kernel: MFR bundle ls-0/0/0:0 link t1-1/0/1
differential 90.0 ms over yellow differential delay 75 ms
Feb 17 21:01:53 /kernel: MFR bundle ls-0/0/0:0 link t1-2/0/1
differential 100.0 ms over yellow differential delay 75 ms
=20
The customer routers are all Juniper J6350; I believe the Carrier's
routers are all Cisco GSRs.
=20
Advanced JTAC says that our configurations are solid and that there are
no known bugs that would exhibit behavior like this. The carrier is
insisting on performing physical-level tests of the circuits (even
though they're running error free) before they'll engage higher-level
engineers so I'm currently in a holding pattern awaiting those results.
=20
My Google-foo is failing me and I'm not able to find any documents that
help explain what may be causing this and how to troubleshoot and find
an eventual solution.
=20
I would really appreciate any tips or suggestions from anyone on the
list that may have seen issues like this in the past.
=20
Thanks,
=20
Ben
=20
=20
