Multilink on Cisco 7513.

[sagar.joshi]

Hi,

One of Customer is using Cisco 7513 having PA-8E Module installed in it. We have 8 E1's. Router is running with 12.0 25 S1. We have configured Multilink on Router. The another end is having Juniper Router. When we use multilink with 3 E1's, it will work fine with no problems. But as soon as we add 4th E1 we are receiving MRRU Exceeded Errors on Juniper End.

What might be the problem ??

Can any one help me out ....

Regards,

Sagar Joshi.

[marikakis]

In the following URL:

http://www.cisco.com/en/US/products/sw/iosswrel/ps5207/products_feature_guide09186a00801ee189.html

you can see some advice about MRRU.

"When using a large-bundle interface MTU size, you must ensure that the individual frames-per-fragment size passed to the link interfaces is not greater than the link interface MTU setting or the peer MRRU setting."

"With MLP the MTU size of the bundle interface must not exceed the MRRU setting of the peer. The MRRU settings on both sides need not be equal, but the "must not exceed" rule just specified must be followed; otherwise a system might send several fragments that, when reconstructed as a frame, will be too large for the peer's receive buffer. "

Try to adjust the MTU setting on the 7513 MLP bundle, and see if that fixes your problem.

M.

[sagar.joshi]

Hi,

Thanks a lot.

Regards,

Sagar Joshi.

[gscholz]

Sorry to butt in Please clarify

I am running into issues with Multilink on a simple bundled connection for BRI 128K connecting to a PRI with latency which appears to be related to the reordering of packets. I am getting 2 second spikes and although these are not long in duration I was wondering if I have to configure something in relation to

* endpoint dicrimination ( the bundle names are the same at both ends)

* the MRRU as I am connecting from a PRI at the Core site or

* the fragment delay time

* will Multichassis PPP Multilink 12.2(13)T solve my problem ( promises without loss of sequencing)

my configuration currently which is a little overkill

no fair-queue

compress stac

ppp multilink

ppp multink fragment-delay 80

ppp multilink interleave

If i go to 12.2(13)T

can I get away with

ppp multink

ppp multilink multiclass

Also what overhead does fragmenetation add

I have moved 4 Primary rates from a 7500 to a 3660 and I am seeing considerable latency difference since the change. Does the backplane have some bearing on this. CPU load does not appear to be a factor

Looking forward to your reply