Mailing List Archive

Route Reflection
Juniper documentation says:

cluster-identifier -- IP address to use as the cluster identifier.

What IP address should be used and what is it used for?

Any pointers would be very helpful.

Thanks,
Jay

--
Jay Young |Office of Information Technology
Network Engineer |The Ohio State University
Phone: (614) 292-7350 |320 West 8th
Fax: (614) 292-9525 |Columbus, OH 43201
Route Reflection [ In reply to ]
Jay,

This can be any ip on the router or it can be some other significant number
in dotted decimal. It is placed in the cluster_id field of reflected route
updates to prevent loops.

Cody


----- Original Message -----
From: "Jay Young" <jay@net.ohio-state.edu>
To: <juniper-nsp@puck.nether.net>
Sent: Thursday, September 19, 2002 12:06 PM
Subject: [j-nsp] Route Reflection


Juniper documentation says:

cluster-identifier -- IP address to use as the cluster identifier.

What IP address should be used and what is it used for?

Any pointers would be very helpful.

Thanks,
Jay

--
Jay Young |Office of Information Technology
Network Engineer |The Ohio State University
Phone: (614) 292-7350 |320 West 8th
Fax: (614) 292-9525 |Columbus, OH 43201


_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
http://puck.nether.net/mailman/listinfo/juniper-nsp
Route Reflection [ In reply to ]
On 19/09/2002 11:06, "Jay Young" <jay@net.ohio-state.edu> wrote:

> Juniper documentation says:
>
> cluster-identifier -- IP address to use as the cluster identifier.
>
> What IP address should be used and what is it used for?

The cluster id is defined as a 4-byte identifier. In our implementation
this is configured as an IP address. You can use any IP address as the
cluster id so long as it is unique per cluster. At no time is it used in
routing so of this you can even use Martians. Like 0.0.0.1

It is only used to identify the cluster so that RRs can discard routes from
other RRs in the same cluster, in the case of redundant RRs and in the
Cluster list which shows the reflection path the route has taken.

> Any pointers would be very helpful.
>
> Thanks,
> Jay

------
Gary
Route Reflection [ In reply to ]
On Thursday, Sep 19, 2002, at 14:38 US/Eastern, Cody DeNiro wrote:

> This can be any ip on the router or it can be some other significant
> number
> in dotted decimal. It is placed in the cluster_id field of reflected
> route
> updates to prevent loops.

Most people use the IP of the router's loopback interface. It doesn't
matter what IP you use, it just has to be unique.

Hope this helps.
- Jason Parsons
Route Reflection [ In reply to ]
> > This can be any ip on the router or it can be some other significant
> > number
> > in dotted decimal. It is placed in the cluster_id field of reflected
> > route
> > updates to prevent loops.
>
> Most people use the IP of the router's loopback interface. It doesn't
> matter what IP you use, it just has to be unique.

It also has to be the same for all route-reflector servers that
"serve" a given set of route-reflector clients. This means that you
either nominate one router of the group of doing the serving to
provide the loopback to use as cluster identifier, or you create a new
space of numbers that don't overlap with your loopbacks (like 0.0.0.1
to identify your first cluster, 0.0.0.2 the second, and so on).

I personally favor the latter for clarity.

Stephen