If implement DFCs, you move from a 30 Mpps centralized forwarding system anywhere up to a 400 Mpps distributed forwarding system HOW??

[mukupatil]

Hi,

Cisco Claims: - Performance is the biggest and most obvious reason to implement DFCs. You move from a 30 Mpps centralized forwarding system anywhere up to a 400 Mpps distributed forwarding system.

I know how we get the 400Mpps but no idea what they are tring to say when copare with CFC.

I dont understand the concept and calculation here...

Regards,

Mukesh

[Reza Sharifi]

Hi,

Can you post the link to the doc you are reading?

[rsimoni]

Hi Mukesh, 

I posted the answer already replying to you post

https://supportforums.cisco.com/message/3541030#3541030

what is not clear?

Riccardo

[Leo Laohoo]

Hi Mukesh, 
I posted the answer already replying to you post
https://supportforums.cisco.com/message/3541030#3541030
what is not clear?

LMFAO  (+5)

[mukupatil]

Cisco URL..

http://www.cisco.com/en/US/products/hw/switches/ps708/products_qanda_item09186a00809a7673.shtml

Statement is under

Q.

What are the benefits of a DFC?

[mukupatil]

Dear Riccardo,

The answer you gave to my previous question on 400Mpps calculation was absolutely clear.

I am confused, when the Cisco says "Performance is the biggest and most obvious reason to implement DFCs. You move from a 30 Mpps centralized forwarding system anywhere up to a 400 Mpps distributed forwarding system."

I am not 100% clear on the statement given above regarding CFC 30Mpps performance. Does it mean if I have LC without the DFC then every packet coming to LC will go to PFC for L2/L3 forwarding and I can not cross the 30Mpps performance as a whole chassis (inc any Non DFC-LC) because SUP can do max 30Mpps in centralized model which result in max performance of 6500 without DFC is just limited to 30Mpps as a whole.

Apologies for my vague understanding

Regards,

Mukesh

.

[rsimoni]

Hi Mukesh,

Does it mean if I have LC without the DFC then every packet coming to LC will go to PFC for L2/L3 forwarding and I can not cross the 30Mpps performance as a whole chassis (inc any Non DFC-LC) because SUP can do max 30Mpps in centralized model which result in max performance of 6500 without DFC is just limited to 30Mpps as a whole.
Apologies for my vague understanding

Your undertsanding is not vague at all!!! you got it 100% right!

CFC cards use the central PFC for L2/L3 lookups, so if no DFC card is available the max performance of a 6500 chassis is 30Mpps (or 48Mpss if you have the Sup card with 10Gbps uplinks).

Riccardo

[Joseph W. Doherty]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

As Riccardo notes, your understanding is correct, but wanted to touch on your statement:

Does it mean if I have LC without the DFC then every packet coming to LC will go to PFC for L2/L3 forwarding 

My understanding, the DFC only process packets that arrive on the LC's ports, not the fabric.  So for instance, if you had two 6748 cards, one with DFC and one without, passing traffic between two ports, one on each card, the one with the DFC would forward the ingress packet directly to the other card but the one without the DFC would rely on the sup to forward its port ingress packets.

[rsimoni · Accepted solution]

Hi Joseph,

well, that is not totally exact. Traffic goes through the fabric even if ingress LC has a DFC.

The difference is that without DFC we have a shared lookup model where all the forwarding decision is taken on the centralized engine (PFC). With DFC the ingress card does not have to send the header (DBUS header after adding platform specific fields) to the PFC and wait for the lookup decision (sent through RBUS) before sending the packet to the egress port via the fabric but sends the header to a local forwarding engine (DFC) which returns the same information returned by the PFC (therefore inclusing the information about how to send the traffic via the fabric .

Packet fowarded locally on the same DFC enabled card follows slightly different path though.

Riccardo

[mukupatil]

Here is the DFC decision

[Joseph W. Doherty]

Disclaimer

The    Author of this posting offers the information contained within this    posting without consideration and with the reader's understanding that    there's no implied or expressed suitability or fitness for any  purpose.   Information provided is for informational purposes only and  should not   be construed as rendering professional advice of any kind.  Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In    no event shall Author be liable for any damages whatsoever  (including,   without limitation, damages for loss of use, data or  profit) arising  out  of the use or inability to use the posting's  information even if  Author  has been advised of the possibility of such  damage.

Posting

Traffic goes through the fabric even if ingress LC has a DFC.

Riccardo, I re-read my post several times to understand why you may have thought I thought to the contrary.  Did you think ". . . one with the DFC would forward the ingress packet directly to the other  card but the one without the DFC would rely on the sup to forward its  port ingress packets." to mean I was discussing packet flow?  If so, my apologies for poor wording as I was attempting to clarify the role of DFC in the forwarding decision; nothing was intended about actual data transfer path.

Actual data transfer path is, I believe, independent  of the DFC, as it relies on what connections the source and destination  LCs have: bus only, bus and fabric, or fabric only connections (and also  whether there's fabric installed).

Again, though, want to emphasis the DFC contributes to the forwarding decision performance for packets/frames received on the LC's (edge) ports, not to packet/frames received from another LC, whether from a fabric (or bus).  (I.e. DFC participation in forwarding is seen in step 3 in Mukesh's recently posted diagram.)

[rsimoni]

Hi Joseph,

no need to apologize as after all the one having a poor understanding is me  thanks for clarifying anyway.

Sorry about that.

Riccardo