[c-nsp] NCS-5501 & 5502 Experiences?

Adrian Minta adrian.minta at gmail.com
Fri Feb 3 08:07:49 EST 2017


Hi Gustav,
How bad NCS5501 not SE variant is oversubscribed ? From the data sheet 
"up to 800 Gbps of system throughput" fall short of the total 48x10 + 6x100.
Does NCS5501 has a LPTS/CoPP implementation similar with ASR9k ?

Many thanks in advance !

On 02/02/2017 10:56 AM, Gustav Ulander wrote:
> Hello.
> We are using NCS 5001 in this role with great success haven't had any issues with it.
> They seem to be just working although that's not that surprising given that its not that complex of an environment.
> We don't do OSPF-TE or RSVP-TE though so on that I cannot comment.
>
> //Gustav
>
>
> -----Original Message-----
> From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of James Jun
> Sent: den 2 februari 2017 00.35
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] NCS-5501 & 5502 Experiences?
>
>
> Is anybody rolling NCS 5501 or 5502 in production yet with the current IOXR 6.1 build?
>
> We're evaluating deploying NCS 5501 and 5502 (the -TR variants, not SE) for some MPLS LSR role.
> If selected, the NCS55K would be used as pure 'P' router where all it does is label switching in the middle, with ASR9K and MX boxes serving as PE's using 100GE into NCS55k.  We do not need L2VPN, L3VPN or any PE services.
>
> Our requirements are fairly simple, as noted below.  I've already run this through with our SE, and he had confirmed that NCS5501 and 5502 should support them all, but I was wondering if anybody had any experience using these boxes in production.
>
>
> Requirements:
> - MPLS Traffic Engineering (using OSPF-TE) and RSVP-TE
> - Fast Reroute for RSVP-TE LSPs
> - Bundle-Ether/LAG hashing - src/dst + ports, labels, for labeled traffic.
>    Majority of labeled traffic expected to traverse through the NCS are typical IP/IMIX internet
>    traffic, but labeled 1 time for PE-to-PE LSPs; IPv6 traffic would see two labels due to 6PE;
>    There is also some limited pseudowire and vpn traffic so hashing on labels would be needed
>    as well.  For the latter case regarding l2vpn, PEs will build l2circuits using FAT-PW.
> - No need for large FIB, just decent LFIB to hold LSPs and FRR bypasses for typical metro size
>    regional network.
>
> - The port buffers on NCS5501 are deemed sufficient enough for 100GE->10GE stepdown on bursty
>    internet use/IMIX traffic, but real life experiences/feedback would be nice to hear.
>
> - We do not need to run BGP on these; out-of-band route-reflectors are used.
>
>
> Many thanks in advance,
>
> James

-- 
Best regards,
Adrian Minta




More information about the cisco-nsp mailing list