foundry-nsp at ergens
May 27, 2012, 1:43 AM
Post #2 of 2
A common method I see some ISP's doing lately is to connect two
switches on each side seperatly with each other.
Something like this:
SA1 ===== SB1
SA2 ===== SB2
No stack or interconnection between the switches in the core at the same site.
Then you install VPLS or simple q-in-q's from each port of each core
switch to the same port on the other core switch in the other site.
This will give you a sort of 'long line' patch panel. Port 1 on site A
is port 1 on site B.
Redundancy is then created on the access layer. You can use for
example spanningstree or dynamic trunks (lacp short). Just make sure
you are using active protocols which can detect the failure of the
line between the datacentres as your access switches will not receive
a link down notice.
2012/4/24 Kurt Blaser <kurt.blaser [at] senselan>:
> We are a small ISP in Switzerland. We have build two Datacenter and want
> to start to interconnect them for our Colocation - Customers and our
> ethernet services. For this project we have 4 Netiron Routers. Our idea
> is to connect the distribution switches (Layer2) redundant to the core
> (two new Netiron routers per site). With Netiron 5.3 the feature MCT
> with VPLS are now avaiable. Is there any experience with this feature?
> Or is good enough to do it with MCT? We want prevent to implement Rapid
> Spanning Tree in this case. We have the advanced software license that
> can do vpls and 802.1ad.
> Or any other ideas?
> Kurt Blaser
> ps Sorry for my poor english
> foundry-nsp mailing list
> foundry-nsp [at] puck
foundry-nsp mailing list
foundry-nsp [at] puck