[c-nsp] ME3800X/ME3600X/ME3600X-24CX/ASR903/ASR901 Deployment Simplification Feedback

Mark Tinka mark.tinka at seacom.mu
Mon Jul 22 15:39:58 EDT 2013


On Sunday, July 21, 2013 11:29:01 PM Waris Sagheer (waris) 
wrote:

> Hi Mark,

Hello Waris.

> - 48 gig port switch requirement, I suppose you also need
> 4x10Gig uplink along with 48 Gig port, correct?

Yes, we'd need at least 4x 10Gbps uplink SPF+ ports for a 
switch that came with 48x Gig-E ports. Of course, the 
assumption is that you could only uplink 40Gbps, but 
assuming all ports can run at line rate, that should not 
limit local switching between Gig-E ports if customers are 
communicating with one another via the same switch.

> - Per
> pop growth, I'll get get back to you with a solution to
> seek your feedback

Many thanks, Waris.

> - Can you give me an idea in terms of
> number of FIB entries requirement?

I think a compromise between cost of the switch vs. where 
the world is going re: IPv4 and IPv6 table growth, I'd be 
happy if Cisco can support 1,000,000 FIB entries (whether 
IPv4, IPv6 or MPLS) in newer hardware provided it doesn't 
break the bank. I think anything more than that and this 
product line quickly begins to lose its appeal as an Access 
IP/MPLS router with high port density, since it will start 
to cost more and we're back to square one.

But like I said, this is not such a real issues for me, 
personally, as this switch line has more important things to 
do like remove STP from the Access rings :-). So even if you 
were to bump current FIB slots by just even double, I'm sure 
someone out there would be happy.

> - Can you elaborate
> your comment "(particularly coming as close to the
> flexibility of what software routers like the 7200 can
> do)"? Any 7200 example functionality.

I was referring to MQC, and how on software routers, it can 
really do anything you want.

On hardware-based platforms, you find certain QoS features 
aren't supported, e.g., when the ME3600X/3800X first 
shipped, there were restrictions on QoS match conditions 
(match-any vs. match-all), queue depth limitations, how 
interface bandwidth %'s were allocated in QoS policy maps, 
e.t.c. Of course, since these QoS configurations get 
programmed into hardware, there is little margin for 
flexibility than when compared to a software-based router; 
so what I mean is, wherever possible, making QoS on the 
ME3600X/3800X as flexible as we have it on the software-
based routers, assuming the hardware can do it without 
adverse side effects on another functionality of the switch.

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/cisco-nsp/attachments/20130722/ee41cbd8/attachment.sig>


More information about the cisco-nsp mailing list