The load balancing on the ae bundle is controlled by the ae configuration and process. The mpls side configuration is all about which links to use and once on that link the load balance is handled normally. Running these on an mpls network the main changes we made were to increase the bfd timer on the links to avoid flapping and to make sure the appropriate changes were made to the bandwidth advertised by rsvp.
The details on ae load balancing in Junos are here.
https://www.juniper.net/documentation/us/en/software/junos/high-availability/topics/topic-map/load-balancing-aggregated-ethernet-interfaces.html------------------------------
Steve Puluka BSEET - Juniper Ambassador
IP Architect - DQE Communications Pittsburgh, PA (Metro Ethernet & ISP)
http://puluka.com/home------------------------------
Original Message:
Sent: 05-10-2022 09:23
From: ADRIAN PROCTOR
Subject: MPLS Core Link Aggregation
Hi,
I am part of a team that supports an MPLS core network comprising of Juniper MX480 PE routers (no P routers). The PE routers are currently running Junos version 19.4R3.11 and use MPC3E-3D-NG-R-B with MIC3-3D-10XGE-SFPP for 10G connectivity. RSVP is used for label distribution and we don't use CSPF. We would like to scale up the 10G MPLS core links using link aggregation (i.e. 10G to 20G or 30G). Does anyone have any experience in doing this? Other than the usual basic LAG/LACP configuration, do I need to consider any extra settings to improve load balancing? Are there any Juniper guides that cover this area in detail?
Many Thanks
Adrian
------------------------------
ADRIAN PROCTOR
------------------------------