[j-nsp] CoS Marking/Rewrite Theory - Update!

[Mark Tinka]

On Thursday, September 02, 2010 07:45:45 am Chris Evans 
wrote:

> I asked for this years ago but was blown off.  Basically
> got the 'that is the way it is' statement.  As stated
> the current implementation is very limiting.

Probably just about the most frustrating quirk on the JUNOS-
based platforms that we continually experience, not 
forgetting the massive hardware restrictions on the line 
cards that call for Services modules on many of the systems 
for even the most basic of features.

It's annoying that we paid lots of money for the top-of-the-
line DPC (DPC-E-Q-R) which touts massive QoS support and 
feature set, but can't even handle the new Translation 
tables.

On the M7i/M10i, for example, we'd be forced to use the 
IQ2(E) PIC if we want Translation tables, which would mean 
we're oversubscribing the slot. Yes, I could only provision 
one port on the PIC for line rate, but then I'm paying for 3 
ports that I'm not using. That said, the M7i/M10i has aged 
enough for me to quiet-down a little :-).

> Cisco has had this feature since I can remember on all of
> their platforms.

I couldn't agree more.

I mean, this is one feature we didn't even have to think 
about until we started using Juniper systems at the edge. I 
wonder why Juniper thought it would be more difficult to NOT 
remark traffic as close to the customers as possible.

Cheers,

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/juniper-nsp/attachments/20100902/d7a2d78a/attachment.bin>