[c-nsp] STP to L2 FTTx gear

[Eric Helm]

Justin Shore wrote:

> We're also the L2 provider, though it's not on our usual Cisco gear.  We 
> chose Occam as our FTTx vendor.  The GigE ring and access layer is made 
> up of Occam switches.  Occam has taken what I'd consider to be an 
> unusual approach for VLAN numbering.  They retag (not QinQ but VLAN ID 
> retagging) our 8xx VLANs based on the VLAN's purpose to what I believe 
> are hardcoded VLAN IDs on their gear.  For example management uses VLAN 
> 2.  Voice uses VLAN 5.  Public Internet data uses 4.  Etc.  This 
> translation should only occur on the upstream edge touching my L3 core. 
>   One thing that I do not know is if my BPDUs are traversing the entire 
> ring before popping out on the other GigE linecard and into my second 
> 7600, or if the BPDUs are passing directly between the 2 GigE linecards 
> in our CO and aren't traversing the ring.  I need to inquire with the 
> vendor about this.
> 

Justin,

I've had some experience with the Occam FTTx gear, and it is a bit odd
the way they handle those internal VLANs, but they do it because they
hard code the QOS based on VLAN ID, so they simply tranlate their
internal VLANs to whatever VLANs you want to use in the rest of your
network. In my past experience with Occam, it was with Foundry L2
aggregation gear, and we chose to use Foundry's Metro Ring Protocol
instead of STP. It worked great in conjunction with Occam's Ethernet
Protection Switching protocol. Although, I've not used Occam with Cisco
gear, Occam's website has a whitepaper that suggests using Cisco's Flex
Link technology for high availability, which would remove the need for STP.
http://www.occamnetworks.com/pdf/CiscoSvcFlxOccam.pdf

/Eric