Hello everyone
Recently I have a network design case for old-large network which currently having problem with STP (in fact the network diameter is too huge for BPDU propagation), as well as VLAN starvation! (it seems their tenant consumed all 409x VLANs)
In new network design we're looking for a solution, which allow each tenant will be able to manage their own network without disturbing the core network. (use their own VLANs, overlap IP addresses, and end-to-end connectivity), I consult with several principles, some of them suggested QinQ, by creating provider VLAN, we can serve separate VLAN for each tenant, and then tenant have their own freedom to use 409x VLANs and IP schema.
However, firstly QinQ is considered as Layer 2 technology, and we still need to have end-to-end (campus wide) provider VLAN across our network. as a result, broadcast still remain in picture. secondly in QinQ, STP again highly involved in core, even upgrading to 10/40 Gbps links in core network, not 100% guarantee that BPDU will reach on time to all layers of network. Additionally we have less flexibility with layer 2 compare to layer 3. in other word problem of the Layer 2 will move to the new network.
Some other principles suggested MPLS. Obviously, MPLS is de facto at the moment; modern technology, and more flexibility compare to QinQ. and personally I prefer Layer 3 and dynamic routing protocol be involved in my core network.
When both QinQ and MPLS considered as routing through core technology, I want to know your opinion regarding of comparison of QinQ and MPLS?
what are the advantages and disadvantages of each technology? and which one do you suggest to choose?
Regards
Siavash
