Re: MPLS / Switching Method on Non Cisco Devices

I guess my message was not clear for you.
(WAN?) encapsulation has all to do with MPLS being hardware friendly
for switching, irrespective of the platform.
CEF is a way to do hardware lookup of a large size routing key
(destination IP) in a hierarchical structure (FIB).
CEF mostly underutilized for MPLS, although the one size fits all
way of using it for IP & MPLS at the same time works ok.
Also, IP is as fast as MPLS from a switching standpoint in cisco's
implementation AFAIK, and that is currently underadvertised because
it shadows one of MPLS's perceived benefits.

Again, my point of view.
-Carlos

Ravi Singh @ 13/01/2010 10:37 -0300 dixit:
> Ahh..a reply finally ;-) .. Thanks Carlos .. however, my query is more
> oriented towards the switching method utilized for MPLS in Non-Cisco routers
> (Juniper etc .) rather than on the WAN encapsulation type or device
> used..for example a cisco router would build a LFIB and LIB using CEF , but
> since non-cisco routers do not use CEF, do they build the same data
> structures and if their packet switching method is similar to CEF which
> builds FIB and adjacency tables..
>
> Thanks for your reply again,
>
> Ravi
>
> On Wed, Jan 13, 2010 at 1:18 PM, Carlos G Mendioroz <tron_at_huapi.ba.ar>wrote:
>
>> Well, it seems that conceptual things are not very attractive as topics.
>> (just some pepper to the mix :)
>>
>> CEF is cisco's way to do it, but in general, MPLS is hardware friendly
>> no matter what way you do it, because it has a fixed size fixed position
>> small label that can be used to determine where the frame/packet should
>> go in constant time.
>> ATM has a even better constant size frame (cell), but at the cost of
>> forcing SAR (segmentation and reassembly) on upper stack.
>> FR/ATM switches usually specify how many *bits* they can read in
>> hardware to switch traffic, I guess nowadays MPLS-able switches should
>> do the whole label.
>>
>> As to who decides the label, there are two alternatives to when it is
>> done but always the link receiver is the one. Frame mode allows for it
>> to be determined up front, usually when the FEC is created.
>> Cell mode does it on-demand, to economize the number of circuits
>> that are mapped to them. (label mapped to VPI/VCI)
>>
>> My understanding at least, hope it helps.
>> -Carlos
>>
>> Ravi Singh @ 6/01/2010 8:52 -0300 dixit:
>> > Greetings Everyone ,
>>> From all the MPLS material I have read, I understand the dependency of
>> CEF
>>> on MPLS . However, just out of interest, I would like to know what
>> switching
>>> methods do the non-Cisco equipment utilize for MPLS functioning. I tried
>>> finding the information on the web but could not get any valuable info.
>>> I know nothing about Juniper or any other non-cisco equipment for that
>>> matter but I am just curious to know how different is the switching
>>> behaviour in these devices when compared to Cisco. So I am really not
>>> looking for detailed technical stuff but just a clarity on who takes the
>>> care of the label assignment in such a case.
>>>
>>> Cheers,
>>> Ravi
>>>
>>>
>>> Blogs and organic groups at http://www.ccie.net
>>>
>>> _______________________________________________________________________
>>> Subscription information may be found at:
>>> http://www.groupstudy.com/list/CCIELab.html
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>> --
>> Carlos G Mendioroz <tron_at_huapi.ba.ar> LW7 EQI Argentina
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>

-- 
Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
Blogs and organic groups at http://www.ccie.net