My questions are: Should I continue to use a single ASN for all bgp peers? currently using '1'
At Tower1, should I use a single instance to both connect to Core and to Towers 2,3,4, or should I have an 'upstream' instance and a 'downstream', or even a different instance for each tower I connect to?
For route filters, should I be doing accepts and only allowing specific networks through? Or should I be letting anything through and disguarding certain networks?
Finally, when to use route-reflect. Just at Tower1 so that Towers2,3, and 4 can get those routes?
Joined: Thu Jan 27, 2011 7:29 am Posts: 186
Karma: 1
just to throw a kink in here, I plan on deploying an HWMP+ mesh on this and giving each tower router a public IP on the mesh and eventually routing clients through the mesh.
When I do that I will have a separate bgp tree on the mesh, but because the mesh is a single layer2 I will just have the core mesh node with route reflect on and the sites as peers, all of which on a separate bgp instance. (unless I am told that is the wrong way to do it.)
Joined: Sat Apr 04, 2009 2:45 am Posts: 579
Karma: 5
Location: Østfold - Norway
run a main ebgp at core router. Then try confederation with different AS in all ibgp routers. make sure u have a connection between all routers, who will act as multihomed (redundant) links. if stright r1-r2-r3-r4 and no r2-r3and4 its no good use of bgp. Just stight forward routes.
Joined: Thu Jan 27, 2011 7:29 am Posts: 186
Karma: 1
so in my example:
Quote:
Core <bgp> Tower1 (tower 1 is a main distribution point to other towers) Tower1 <bgp> Tower2 Tower1 <bgp> Tower3 Tower1 <bgp> Tower4
I can extend this to say Tower2 connects to Tower 3, which makes Towers 1,2,3 in a triangle shape. Tower3 connects to Tower 4, making 1,3,4 a triangle shape.
This means a ton for static routes.
Also, if I have a subnet at the end of this topology Core>Tower1>Tower4>Tower5>CPE, each tower has it's own subnet which means 4 networks, I have 4 static routes to add ibgp means 0 static routes.
I was thinking I would use ibgp for this for simplicity, but without route filters I am getting huge route lists and loops etc etc.
Joined: Fri Jan 30, 2009 11:41 am Posts: 139
Karma: 14
Location: Krakow, Poland
syadnom wrote:
Maybe I should be looking at OSPF instead?
And how many "real" eBGP sessions (on different routers) do you have there? iBGP's main idea is that the eBGP routers should "discuss" the routes received from their peers and decide, how to route the traffic outside. For routing inside one "real" AS - BGP is too "heavy" and complicated, and THAT'S WHY people invented some better routing protocols - like OSPF!
_________________ Did this post help you somehow? Don't forget to give karma
Users browsing this forum: No registered users and 2 guests
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot post attachments in this forum