This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
I'm attempting to stack / use MLAG with two M5300-28G3 switches (both same version 11 firmware). I have two AX742 stacking modules on each switch (so 4 total).
I have a few conceptual questions before I start, as the manual isn`t totally clear to me on those points. I have never done any stacking/MLAG, but most of the instructions seem clear.
1) When stacking more than two switches, I understand why a ring topology is best, but when stacking only two switches, do you need 4 stacking modules to make it a ring or will only one link (two modules) be sufficient? In principle it would seem with only two switches the ring topology is not adding anything.
2) Do you need the switches to be stacked when doing MLAG? (this I`m pretty sure is a "yes" but the manual seems to take for granted I would know that)
3) The manual mentions a very important peer link (a LAG between both switches that will do MLAG) to do MLAG. But if those switches are stacked using AX742 modules, do you also need a peer link or does is this stacking link sufficient?
4) Let say I need a "stack" link AND a peer link (question above), would best practice be using the AX742 modules as stack links and the 10G Ethernet ports as peer link, or vice versa? Why?
That's great. Redundancy is actually important and it will be depending on how many switches are you going to deploy. It might be better if you may just use the stacking links as LAG or what we call as distributed LAG. I suggest you also to have a 2nd uplink from another switch and let the RSTP do their thing.
1) When stacking more than two switches, I understand why a ring topology is best, but when stacking only two switches, do you need 4 stacking modules to make it a ring or will only one link (two modules) be sufficient? In principle it would seem with only two switches the ring topology is not adding anything.
- For stacking 2 switches, it would be more efficient if you will be using only 1 link. It will depend on you if you wanted to add a backup just in case the other port goes down.
2) Do you need the switches to be stacked when doing MLAG? (this I`m pretty sure is a "yes" but the manual seems to take for granted I would know that)
- Well, it's not really needed but it will always be on your own preferences. Stacking makes it easy for you to configure all the switches that you having inside your network with one management IP.
3) The manual mentions a very important peer link (a LAG between both switches that will do MLAG) to do MLAG. But if those switches are stacked using AX742 modules, do you also need a peer link or does is this stacking link sufficient?
- NETGEAR switches may call it as Distributed LAG. When you set the stack links as LAG. The stacking link is sufficient.
4) Let say I need a "stack" link AND a peer link (question above), would best practice be using the AX742 modules as stack links and the 10G Ethernet ports as peer link, or vice versa? Why?
- As I've mentioned above, it will always be depending on how your network to be setup. You may use it freely. Peer links is just a back-up.
Thank you John - I think I was confused with MLAGs vs Stacking. I understand they are slightly different even if I don`t understand everything about the differences themselves, but I also understand that in my situation they both provide me with redundancy and stacking with distributed LAG is easier. I am not even sure at this point if MLAG is supported by the M5300.
Thanks also for clarifying that I don't strictly need 2 stack links for stacking only 2 switches, but having two allows to redundancy in case of port/cable failure. We`ll be going with redundancy in our case since it would be pretty dumb to make our switches redundant but not the actual stack link. My understanding of a stack link failure while both switches stay active is that it`s pretty disastrous for LAG network activity.
All in all the M5300s seems to be a very nice solution, and few failure tests we've run are all satisfactory.
That's great. Redundancy is actually important and it will be depending on how many switches are you going to deploy. It might be better if you may just use the stacking links as LAG or what we call as distributed LAG. I suggest you also to have a 2nd uplink from another switch and let the RSTP do their thing.
