[c-nsp] NSE-150 versus NPE-G100 versus NPE-G2

[Saku Ytti]

On (2007-03-21 11:41 -0400), Aaron Daubman wrote:

> I'm trying to gather information to make an informed decision between
> purchasing some of Cisco's newer router options.  I'm having
> difficulty finding good information to compare the:
> 7200VXR with NPE-G2

This has somewhat immature software today, but I'm sure it'll
mature up fast. It most likely has long live ahead of it, you
can get gray PA's dirty cheap to it. It's specced for 2Mpps,
but every feature will reduce it's performance. I would 
personally think twice to deploy it even setup with 1Mpps,
depending on how much QoS, ACL etc will be used.

> versus
> 7304 with NSE-150

This is NPU based box (like most modern boxes in fact seem to do). 
It's specced for 3Mpps, but things like L3 MPLS VPN require two
lookups and ends up being 1.75Mpps. It might make sense if you
have lot of QoS or ACL, don't need IPv6 and can't afford 7600.
I don't think 7304 has long life ahead of it, and I wouldn't
recommend it, I'd go from NPE-G2 to 7600.

> versus
> 7304 with NPE-G100

This product makes little sense for any applications, perhaps if you
have to have software box with redundant control-plane. It's basically
NPE-G1, only you need to pay lot more for interfaces of at very least
buy PA/CC adapter.
If you don't absolutely need redundant RP and can't pay for 7600,
stay away from this.

> These routers will be used in a mostly IPv6 R&D environment where IPv6
> and GigE performance are the most important requirements.  Perhaps my
> cisco-search-fu is just lacking, but I couldn't find much detail about
> NSEs/PXF and IPv6 performance.

Forget NSE100 and NSE150 then, they do not do IPv6 in PXF (NPU) but
in software in ~NPE300 performance. NSE150 might get IPv6 in future,
but it's not committed AFAIK.

> e.g. GRE is stated as being limited to IPv4 for PXF processing, but
> does that mean that traffic for a  tunnel config like this would also
> get punted to RP:
> 
> interface Tunnel1
>  description native v4-over-v6 tunnel
>  ip address 10.72.10.1 255.255.255.0
>  tunnel source GigabitEthernet0/2
>  tunnel destination 2001:1:1:10::2
>  tunnel mode ipv6
> 
> It's not GRE, but it is IPv6...
> 
> What would be the best way to tell whether the beefier RP of the G-100
> would be better suited to our v6 needs than the application-specific
> PXF?  Is the G2 actually even beefier than the G-100?

Yes, twice.

> (G2 claims 2Mpps, 1.8Gbps backplane.
> G100 claims "more than 1 Mpps", 4.0 Gbps backplane...
> I couldn't find NSE-150 pps claims for RP traffic, just that it was
> "greater than the NSE-100)

Who well NSE100 <300kpps, NSE150 isn't considerably faster I'm sure,
in RP. PXF (NPU) on both is 3Mpps for single lookup.

> 
> Thanks,
>      ~Aaron
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/

-- 
  ++ytti