[j-nsp] MX204 scale/performance

[Mark Tinka]

On 1/16/22 12:00, Robert Hass via juniper-nsp wrote:

> Hi
>
> I'm stating network reorganization moving our infrastructure from
> Cisco machines to bunch of MX204 routers.
>
> I have few questions related to MX204 scale and performance:
>
> 1) How big ACLs are supported on MX204 ? (number of entries for all
> ACLs configured).
> 2) I want to use lt- interfaces to address routing between default
> (global) instance and logical-router. What performance (for IMIX
> traffic) is supported for LT interfaces at MX204 ?
> 3) What FIB programming speed  I can expect from MX204 (number of
> routes per second - new BGP routes, withdraws)
> 4) Can I somehow prioritise which routes are programmed first in FIB ?
> (based on BGP community or policy-statement)
> 5) How many flows for IPFIX are supported on MX204 ? Is IPFIX somehow
> hardware accelerated or done pure in Intel CPU ?
> 6) Any recommendations for JunOS release ? My plan is to use latest
> 20.4, but maybe I'm wrong. Features used: Logical Routers, ISIS, a lot
> of BGP, a lot of ACLe, RE ACL, uRPF, IPFIX

The MX204 is basically the MPC7E line card covered by some metal.

We are running 20.4R3 on ours. Not unhappy.

Mark.