Redundant fibre ring/ethernet ring with CRS

RouterOS General
1

I want to build a ring topology with dark fibers which connects 3 CRS switches using SFP.
So far so good, if I build bridges with RSTP turned on. BUT it is slow in converting to new redundant routes…

Is there a possibility to use carrier grade ethernet oder SPB protocol meanwhile? Is there another trick to avoid the old RSTP?


regards, Jan

2

You could use passive WDM optics with add/drop muxes so the topology is physically a ring, but logically it can be hub and spoke. This gives you the best of both worlds because the physical interface at each router can be a dedicated layer3 interface with no bridge topology to consider. If you want transparent l2 bridging, use VPLS on top of the IP topology.

Here’s a white paper on the subject of the WDM optics.
These are nice because if you have dark fiber, then you can use it for lots of links at the same time - more bang for your buck.
(see figure 4)
http://www.cisco.com/c/en/us/products/collateral/interfaces-modules/cwdm-transceiver-modules/product_data_sheet09186a00801a557c.html

3

Thanks for your help!

Maybe I didn’t get the point using VPLS. How does this help? I can set up a Layer 3 IP structure and use VPLS tunnels from each device to the others. But I need them to switch to the other route if some link is broken! So if I bind the VPLS interfaces to my internal bridge there is also only RSTP to realize the layer 2 redundancy?

4

Not quite - If you’re using VPLS on top of MPLS, then the routing of the underlying IP network provides the redundancy. The VPLS just rides whatever the current best IP pathway is. The bridge should be using split horizon and not RSTP when using VPLS.

So assume you have a bunch of /30 IP links in your topology, and each router has a /32 loopback IP.
OSPF will determine the best path between all of the /32 addresses. If a link goes down, then OSPF will re-calculate the paths and converge on a different route to reach each destination.

Meanwhile VPLS is simply saying “router1 is sending a bridged frame to router4” and it will notice that it now needs to use a new label switch path to reach router 4. It will just start using the new label. When the broken link is fixed, OSPF will re-converge onto the original path, and VPLS will just start using the new label for the original path.

It will all be transparent to the hosts inside the VPLS bridge. (except for some frame drops while the fault is being detected and OSPF is re-converging)

You can speed up the converge times by using BFD in conjunction with OSPF.