[160060] in North American Network Operators' Group
Re: Juniper MX10 and dual stack BGP
daemon@ATHENA.MIT.EDU (David Miller)
Wed Jan 30 21:07:32 2013
Date: Wed, 30 Jan 2013 21:06:09 -0500
From: David Miller <dmiller@tiggee.com>
To: nanog@nanog.org
In-Reply-To: <Pine.LNX.4.64.1301301714190.21401@whammy.cluebyfour.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On 1/30/2013 5:16 PM, Justin M. Streiner wrote:
> On Wed, 30 Jan 2013, Christopher Rogers wrote:
>
>> Does anyone have any sort of performance numbers for the jnpr MX10 series
>> running dual stack ipv4/ipv6? I'm specifically interested in how many
>> BGP
>> prefixes it can handle in dual stacked mode. I've got an environment
>> currently taking 4 full ipv4 tables and a smattering of prefixes coming
>> from a public peering exchange, and I'm curious what will happen if we
>> move
>> to dual stack...
>
> I don't have MXs at the border at this point, but I do have a pair of
> M120s taking 3 full v4 and v6 BGP feeds, plus a few non-transit peers,
> and they're handling the load just fine. Just as a point of reference...
>
> jms
>
According to Juniper, the MX uses separate memory for v4 and v6.
The numbers that I have seen for MX80 are:
v4 FIB 1mil
v4 RIB 4mil
v6 FIB 512k
v6 RIB 3mil
MX10 should be the same as it has the same RE, but just enables/licenses
different #s of ports. These are "conservative" numbers and I have seen
claims of higher actual capacity from some sources.
... I wouldn't expect that you would have any route memory issues moving
to dual stack ...
-DMM
