I have some queries related to Nexus .

[namish.sharma]

I have some queries regarding Nexus Architecture.

1) We have MSFC and PFC for layer 3 & layer 2 communication in 6500 .

How layer 2 & 3 traffic is been handled in Nexus. Do they have a specific asics. What are those asics ?

2) We have three types of line cards in 6500. Classic /fabric enables/fabric only.

Are all line cards in Nexuas are facbric only. What is the Asic which we use for distributed forwarding in line cards in Nexus as we use DFC in 6500.

Fabric modules in line cards are used for providing wire speed or it has other benefits as well. can i say it is just a kind of constellation bus which we had in 6500.

3) Are there and D bus/R Bus / EOBC on Nexus.

I believ D- bus wont be there as it use fabric. What about the results. Does it use the same LTL index logic to calculate result. I believe it should as it is from same family as it is known as EARL 8.

How does it handle control plane traffic / data and result .

Lets say I have a machine A connected to Module 2  port 1 ( M1 card)  and machine B connected to module 3 port 1 on same chassis(7018) .

What will be the whole process which it will follow. Let say layer 2 traffic will be handles by Superman and layer 3 will be handles by Tycho.

if any one can explain me the whole packet flow that would be  great.

4) What is the role of Supervisor in Nexus. It has central Arbitar Asic which id some thing diffrent then 6500. VOQ is also handled here. Can we say if we use central Arbitar Asic on line cards ...They will start working as supervisor some what .

Any comments will be appreciated.

Thanks

Namish

3)

[Marwan ALshawi]

1- Nexus used distributed architecture and all the line cards support L3 except the F1 which you need to mix it with M series to get L3 functionality

2- nexus has diffrnet architecure form 65K however new line cards can provide you line speed 10G per port such as F2

3- Nexus uses fabric modules with full mesh channels between these modules and each line card ( provide you much more capacity than 65K for sure )

for packet flow you can find it on cisco.com because it long a bit however in brief:

- Unicast Forwarding Distribution Manager (UFDM) interfaces

between URIBs on supervisor and IP FIB on

I/O modules

- IP FIB process programs forwarding engine hardware on I/O

modules

4- the Unicast Forwarding Distribution Manager and L2 forwarding manager are examples of what is hosted in the SUP

hope this help