Switching

Hi,

The best practices for EX (2200/4200 etc.) VC state that uplink's LAG member interfaces should not be placed on master or backup RE. What is the reasoning behind this advice?

Regards,

Pawel Mazurkiewicz

This is so that that if you lose a member that is the RE and has an uplink, you do not lose the uplink and the RE at the same time, and if you lose the RE, the other members still have acces via the uplink to the other RE. The attached doc also explains a little bit more

Attachment(s)

VC-Bestpractice-guide.pdf   4.12 MB 1 version

What I do not understand is why is it so important not to loose RE and uplink at the same time? I think that in that case RE would switch to backup, and uplink would work using another LAG member or some other form of link redundancy (RTG, STP etc.)?

> if you lose the RE, the other members still have acces via the uplink to the other RE

That part I do not understand. Other members should access the other RE though VC ports, and not through uplinks to the upper devices (north) I guess...

Regards,

Pawel

You will lose half of your bandwidth.

I wil loose half of the bandwidth regardless of whether or not I place it on the RE. Its just the same if it is in the RE or on the LC...

VC can be connected in two ways: Chanined or Ring. It does not have to be ring. The scenario would be applicable to chained method using the uplink ports. If the uplink ports were on a line card and the that line card failed or the link failed, the devices on that side of the VC would still have access to the RE on that side. If the uplink was on the RE and the RE failed, then they would lose access to the other RE because both the RE and the uplink would have failed. Juniper is always seeking to provide HA in catastrophic failures. Please take a closer look at the topic "using uplinks" as it explains the extended virtual chassis configuration.