with the new 48-port switch it would not be bad to introduce MLAG
- 1
MikroTik Cloud Router Switch CRS354-48G-4S+2Q+RM
+1 but MLAG / LACP must be >20pcs , the best without limitation
and STACKING feature up to 8 units !
I would much rather see stacking.
MLAG is fine for Layer 2, but it’s a nightmare for Layer 3. Stacking works well for both. Considering a lot of the CRS3XX chipsets have L3 HW offload that has yet to be taken advantage of, it would be nice to be able to form an LACP chanel across two or more switches and use it at L2 or L3.
Listen, any solution is fine
but I can’t see netgear in datacenters just because it supports this features !!!
I’d be happy with the ability to just be able to run an MLAG. There is currently no way for one to create a redundant LAG across 2 switches for redundancy.
Anyone have a workaround to be able to bond 2 connections from say a router to 2 switches?
1+ for MLAG
This is hardly a “new” feature request 
Stacking and MLAG have different use cases.
For Enterprise a stack is better
For Datacentre/ISP a MLAG cluster is better
This is primarily down to L3 support and failure domains.
- MLAG
+1
MLAG / Stack whatever but just something or even better both 
+1 for MLAG… this has been talked about for a while and I saw a post ~1yr ago that you guys were trying to get this implemented.
+1 mlag
+1 for both
If the MLAG is active-passive (like in IOS-XR), then Layer-3 is very easy, because only one of both routers has an active/running interface. The distribution needs to be done by a routing protocol, of course.
Edit:
I wonder if it would be possible to fake active passive MLAG using VRRP between both routers for Master election (including a separate HA-link to prevent split-brain). Then employ a script that checks the VRRP-Status (Master or Slave) and accordingly brings up or down the local bonding interface.
Of course the chassis-id needs to be the same on both routers.
The only downside is, that mikrotik forces you to include at least two interfaces in one bond. But you can always use an EoIP-Dummy-interface to circumvent that.
The IEEE 802.1BR support added to 6.48beta12 will eventually give us that, and more:
*) crs3xx - added initial Bridge Port Extender support (CLI only);
*) crs3xx - added initial Controlling Bridge support for CRS317, CRS309, CRS312, CRS326-24S+2Q+ and CRS354 devices (CLI only);
A full IEEE 802.1BR implementation will give you a single control plane for all your switches, like a Cisco 3x00 switch stack or 6800 & Nexus with fabric extenders.
Bring it on, MikroTik!
802.1BR is NOT MC-LAG and has a different use case. It is for bridge extension, port expanders and the like. You could possibly use it for stacking…
MC-LAG is still needed, it removes the “single control plane” point of failure.
It only eliminates the need to use STP and VRRP. Dual control plane issues like lack of synced connection tracking will remain.
I didn’t say it was the same. I said that it probably would bring it with it, and I stand by that. It makes it possible to have LAGs with members connected to different physical chassis. And stacking of some sort (Cisco StackWise Virtual, or VSS. Not NX-OS VPC) is exactly what I want. If a box looks like a single device from the outside, I want to configure it as a single device. Which is why I don’t like Nexus VPC. Or simple ‘MLAG’, for that matter.
Stacking (or whatever you want to call it), when done right, fixes that for you by hiding that you are talking to multiple boxes, unless you specifically look for it.
Or to have 2 or 4Ge redundant connection from server for free.
Anyone tried this yet?
https://help.mikrotik.com/docs/display/ROS/Controller+Bridge+and+Port+Extender
And managed to get a bonded interface across 2 switches?