[alcatel-nsp] BGP mutipath, load balance 7750

[Dejan Tepic]

Hello Mikael,

Interesting about Tele2 and pre-pend. It is not intentional on our side. Also…we are receiving more traffic from Tele2. This is what I’m trying to “correct”.
I was hoping that by balancing outgoing traffic it would affect incoming to some degree.

Thanks for the input. I look into it.

Regards
Dejan


Från: mikael sundqvist <mst at nirvana.se>
Skickat: den 13 mars 2020 13:09
Till: Dejan Tepic <dejan at karlsborgsenergi.se>
Ämne: Re: [alcatel-nsp] BGP mutipath, load balance 7750

Dejan,
If you use a policy to set the origin on both default routes to the same value (igp,egp or incomplete) you will get load balancing for outgoing traffic in that R1 will prefer it’s local peer while R2 will prefer it’s local peer.
If you would like R1 (and R2) to use both uplinks you could turn on ecmp 2 and bgp eibgp-loadbalance in the vprn on both nodes.
This will load balance in the upstream direction but doesn’t change anything for downstream.
I had a quick look at some bgp looking glasses and it looks like you are connected to IP-Only and Tele2, and while IPO announces your /20 prefix as is, Tele2 was announcing it with a ”pre-pend” but I have no clue if this is intentional or not, or if it was just on the looking glasses I looked and this specific prefix. Anyway, to influence downstream traffic you would need another method …

Hth,
//Mikael

From: https://lg.he.net
[cid:image002.png at 01D5F939.D34A1F40]


12 mars 2020 kl. 20:41 skrev Dejan Tepic <dejan at karlsborgsenergi.se<mailto:dejan at karlsborgsenergi.se>>:

Hi Scott and thanks for you input

I dont know why we only take def. routes but nevertheless this is scenario I have right now.
I already have ibgp between R1 and R2 and only outgoing route used is one to R2. Tiebreak reason is origin. iBGP vs incomplete I know I can level the playing field with import policy and set origin to iBGP on both.

I would like to try bgp-multipath 2 with ECMP 2 but im not sure how to go about.
Looking in route table av vprn xx on R1 I see def. route as tunnelled (BGP-VPN) coming from R2.
Do I enable this on both routers with same parametars or do I need to add some special commands as eibgp-multipath or ibgp-multipath

/Dejan

-----Ursprungligt meddelande-----
Från: alcatel-nsp <alcatel-nsp-bounces at puck.nether.net<mailto:alcatel-nsp-bounces at puck.nether.net>> För Scott Weeks
Skickat: den 12 mars 2020 19:30
Till: alcatel-nsp at puck.nether.net<mailto:alcatel-nsp at puck.nether.net>
Ämne: Re: [alcatel-nsp] BGP mutipath, load balance 7750



--- dejan at karlsborgsenergi.se<mailto:dejan at karlsborgsenergi.se> wrote:
From: Dejan Tepic <dejan at karlsborgsenergi.se<mailto:dejan at karlsborgsenergi.se>>

I'm  trying to accomplish load-balance over BGP in multihomed stub scenario.

We are running two SR7750 connected to two different transit providers. Only default routes are coming from transit providers.

ISP A --- R1
               |
               |
ISP B --- R2

BGP with ISP's is done within VPRN. Today, all outgoing traffic is going through ISP B

I would like to be able to load balance outgoing traffic across both ISP's and by doing that influence incoming traffic to be more evenly distributed
----------------------------------------------



Why are you only taking default routes?  If you took full routes with a default route, too, you could accomplish this better.
You'd do iBGP between R1 and R2 along with the existing eBGP you have to each provider.  Then you could use various methods to direct the inbound/outbound traffic in the manner you want.

scott





























_______________________________________________
alcatel-nsp mailing list
alcatel-nsp at puck.nether.net<mailto:alcatel-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/alcatel-nsp


_______________________________________________
alcatel-nsp mailing list
alcatel-nsp at puck.nether.net<mailto:alcatel-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/alcatel-nsp
_______________________________________________
alcatel-nsp mailing list
alcatel-nsp at puck.nether.net<mailto:alcatel-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/alcatel-nsp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/alcatel-nsp/attachments/20200313/6f683b56/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 213459 bytes
Desc: image002.png
URL: <https://puck.nether.net/pipermail/alcatel-nsp/attachments/20200313/6f683b56/attachment-0001.png>