Expanding MS210 Stack Throughput With LACP

[jay-junkbrands]

Hi all, I'm struggling to find clear info on this and I was hoping you could weigh in on my situation.

I just got in two MS210 switches which I've made into a physical stack, and connected the stack to our existing MX84 via SFP. I know that the MS210's can do link bonding via LACP, and I was hoping to bond two SFP uplinks from the stack to the MX84 to make a total 2Gbps throughput, but when I attempt this it takes the switches offline. Is this even possible, or am I misunderstanding?

If not possible, it would be nice to at least have a line on each switch for failover reasons. Would that be possible?

Thanks,

Jay

[Nash · Accepted solution]

As discussed in this thread over here, MX don't do LACP. I'd suggest looking at that discussion.

[jay-junkbrands]

Thanks for the direction, Nash. Looking through that thread clarifies that I can't aggregate.

Do you know if it would work as a failover technique to just have one non-aggregated uplink on each MS210? I understand that only one would work at a time, but if one switch were to go down, in theory the other would activate the redundant line, correct?

[ww]

yes run 2 separate lines from 1 switchstack to the mx works. make sure rstp is enabled on the switch ports 

[GIdenJoe]

Normally the MX is equipped with the correct downlink ports according to the maximum throughput so the need for downstream LACP is not needed.

However what I do find a little problem is the following:
The lack to designate 2 ports to the same WAN link.
Take following scenario into account:

You have two rooms where you terminate a WAN line and there is a router in each room, downstream those routers do an FHRP to have the same subnet available on both locations for redundancy.  So you need WAN1 of your MX available in both rooms, so you connect the WAN1 to a switch stack in that same room and then that switchstack uplinks to both core switches.  However if a switch in that stack fails that happens to contain your only WAN1 link from the MX, the entire device has to do a failover.  Whereas if you would have been able to have a bundle upstream you could connect to two switches in that stack and have better redundancy that way.