0

In following video, the CTO of Arista claims that they use Multi Chassis Link Aggregation (MCLAG) setup for In Service Software Upgrade (ISSU) with zero down time on their switches

https://www.youtube.com/watch?v=VdJZq4dRjf4

Specifically, the claim is from 19.30s to 19.58s

Anyone knows how this is done? There are many network vendors who threw lot of resources in building ISSU and looks like Arista achieved two goals by developing MCLAG feature.

5

I think he's referring to "zero downtime" if you're connected to multiple switches. If you're only connected to Switch A, and I reload that switch for a software update, you're obviously going to lose connectivity. However, with multi-chassis-LACP, you have multiple connections to the "fabric", so a reload of a single switch will not cause a drop.

This is not unique to Arista. Many vendors have this capability.

3
  • This is correct. From the perspective of a device connected to a downstream MLAG pair the logical channel will remain up, even if half of the constituent physical links go down. – rnxrx Nov 10 '17 at 2:13
  • Also, in order to achieve "zero" or "near-zero" traffic loss, there is a dependence on the capability of the downstream switch (which may be from a different vendor). When one chassis of the MLAG pair is being reloaded as part of the ISSU, the links go down, and it is up to the downstream switch to quickly notice the "link down" and quickly switch all traffic on to the surviving link of the LAG (i.e. towards the MLAG chassis that is still up). – mere3ortal Nov 10 '17 at 7:35
  • That's easily handled by the reloading switch removing the link from the group prior to shutting down the link. This isn't a copy-and-power-cycle process. And almost all multi-chassis technologies are vendor specific. – Ricky Nov 10 '17 at 21:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.