topic RE: In switch LAG , 1 port taking more traffic than others. in ExtremeSwitching (EXOS/Switch Engine)

In switch LAG , 1 port taking more traffic than others.

Saurabh_Makkar — Mon, 10 Jul 2017 15:15:00 GMT

We have extreme inter switch with 7 ports in LAG all in different slots.
And we can see 1 of the 7 ports is taking high utilization apart from other 6

As these on different slots, so any input on the issue.
If anything can be checked, (no errors/no congestion) but in peak time sometimes we get ping drop.

show port 1:4 1:5 2:13-15 2:19-20

Port Link Link Rx Peak Rx Tx Peak Tx

State Speed % bandwidth % bandwidth % bandwidth % bandwidth

================================================================================

JBCIMS-U> A 1000 16.35 36.00 29.97 69.66

JBCIMS-b> A 1000 15.42 36.51 14.90 35.03

EVO_Cont> A 1000 31.43 74.81 14.30 32.22

RNCVKIRJ> A 1000 15.50 36.19 14.54 32.33

To_JAISW> A 1000 15.53 40.01 14.61 35.58

RJGMSC5_> A 1000 15.93 40.32 14.81 34.99

RJGMSC4_> A 1000 16.37 40.78 14.88 36.04

RE: In switch LAG , 1 port taking more traffic than others.

Kawawa — Mon, 10 Jul 2017 18:31:00 GMT

It really depends on the characteristics of your traffic and the hashing that is being used when determining what link to forward a stream across. Are you able to confirm what algorithm you're currently using and whether the traffic traversing the LAG is from a specific device. LAGs are primarily designed to aggregate multiple links as opposed to provide load balancing. Therefore if you have (say) the same two devices talking across the LAG, with an L2 and L3 algorithm, these streams will most likely be forwarded over the same LAG port.

RE: In switch LAG , 1 port taking more traffic than others.

BrandonC — Mon, 10 Jul 2017 18:58:00 GMT

Hi Saurabh,

In addition to what Kawawa posted, you may want to take a look at the following GTAC Knowledge article:
https://gtacknowledge.extremenetworks.com/articles/Q_A/Why-is-traffic-distribution-across-ports-in-a...