[j-nsp] How are max routes calculated on an SRX

Skeeve Stevens skeeve+junipernsp at eintellego.net
Wed Oct 17 20:59:49 EDT 2012


Exactly the detailed answer I was looking for and what I thought it should
be.
*

*
*Skeeve Stevens, CEO - *eintellego Pty Ltd
skeeve at eintellego.net ; www.eintellego.net

Phone: 1300 753 383; Cell +61 (0)414 753 383 ; skype://skeeve

facebook.com/eintellego ;  <http://twitter.com/networkceoau>
linkedin.com/in/skeeve

twitter.com/networkceoau ; blog: www.network-ceo.net

The Experts Who The Experts Call
Juniper - Cisco – IBM - Cloud



On Thu, Oct 18, 2012 at 11:38 AM, Ben Dale <bdale at comlinx.com.au> wrote:

> Hi Skeeve,
>
> On 18/10/2012, at 10:04 AM, Skeeve Stevens <
> skeeve+junipernsp at eintellego.net> wrote:
> > Question is.... the specs say it can support a maximum of 700k routes.
> >
> > Now, if what I suspect is right, that is great, but I am getting
> > conflicting information.
> >
>
> The way I read the data sheets is the maximum "supported/tested" in
> forwarding table is 700k.
>
> > A full feed at the moment is about 430k routes.  So I am hoping that the
> > SRX550 could take multiple full feeds... lets say - 3-4, and it just
> > installs a single unique copy of the full feed which is way under the
> 700k
> > threshold.
> >
>
> That is correct.  The advertised values are referring to supported/tested
> FIB, whereas RIB will hold as much as you have memory available (within
> reason).
>
> > But someone was trying to tell me that it couldn't take two world feeds
> as
> > together they are 860k routes.. but my gut feeling is that this is wrong
> as
> > I am not sure what purpose the SRX550 having 700k route capacity would be
> > if it operated that way.
>
> This is simply not true - I have a customer whose SRX100H is taking one
> full view (though data sheet says 8K), and another with an SRX240H with 3
> separate feeds and no issues.
>
> Where you will come unstuck is if you start dumping each feed into
> separate VRFs, then you'll start burning more FIB, and as always the
> infamous KRT queueing "feature" can bite you during fail-over (though I
> have to say I don't see it on SRX/J all that often).
>
> Here's what a J6350 with 2GB of memory looks like with 3 feeds (this is an
> old snapshot back when the DFZ was ~350k):
>
> Groups: 2 Peers: 3 Down peers: 0
> Table          Tot Paths  Act Paths Suppressed    History Damp State
>  Pending
> inet.0           1056579     354871          0          0          0
>    0
> Peer                     AS      InPkt     OutPkt    OutQ   Flaps Last
> Up/Dwn State|#Active/Received/Accepted/Damped...
> x.x.x.x         yyyyy     322391       3804       0       6     1:24:31
> 143268/353090/353090/0 0/0/0/0
> x.x.x.x         yyyyy      86256        192       0       2     1:24:32
> 2/353091/353091/0    0/0/0/0
> x.x.x.x         yyyyy     125597      13456       0       0  4d 4:56:06
> 211601/350398/350398/0 0/0/0/0
>
> Only "active" routes are installed in FIB - the rest sit in RIB.  Here's
> what memory usage looks like for the above:
>
> > show chassis routing-engine
> Routing Engine status:
>     Temperature                 26 degrees C / 78 degrees F
>     CPU temperature             45 degrees C / 113 degrees F
>     Total memory              2048 MB Max  1044 MB used ( 51 percent)
>       Control plane memory    1472 MB Max   618 MB used ( 42 percent)
> <--- RIB sits here
>       Data plane memory        576 MB Max   420 MB used ( 73 percent)
> <--- FIB sits here
>     CPU utilization:
>       User                       0 percent
>       Real-time threads          9 percent
>       Kernel                     0 percent
>       Idle                      91 percent
>     Model                          RE-J6350-3400
>
> Also, when using SRX, make sure all your peers are in the same security
> zone to handle asymmetry issues that may arise.
>
> Ben


More information about the juniper-nsp mailing list