Hi,
I am using two wan links from ISP1 and ISP2 with a /30 wan pool from each upstream provider. Also i am having a /22 IP pool from APNIC. My query is I want to use both of the links simultaneously, two pools of /24 from ISP1 and other two /24 pools from other isp2, in case any of the ISP fails, the pools assoicated with that isp should run through the other ISP link.
All four /24 IP pools are adverstised on both ISP's and I have also tried prepending them by taking help from other blogs but didnt worked.
Can someone assist me in that particular case.

Best Regards
avdhesh

One way that's guaranteed to work is to only announce the /24 prefixes on the preferred ISP. If the preferred ISP goes down, the /22 announcement on the backup ISP will ensure that traffic for those prefixes will be re-routed.

This is a full primary/backup behavior (no load sharing for any given /24)

Of course your outbound traffic will follow whatever path BGP specifies - if you're receiving full routes, then you can tweak this with local_pref based on various destinations.
If default-only, then you'll need to do policy routing.

The best method of ingress traffic engineering is to send communities to your ISPs which govern their route selection / advertisement behavior when repeating your prefixes to their other peers. Ask your ISPs to furnish you with documentation on their published communities and what behaviors they dictate.

You can also prepend the /24 prefixes you want less per ISP.

You can also prepend the /24 prefixes you want less per ISP.

This is the more commonly-applied solution, but it's not guaranteed to work in all cases. If the peer applies a local_pref to your announcements which is superior to any other path it may hear your prefixes on - well, then the ISP itself is going to prefer you no matter how many times you prepend the AS. Most transit providers have a policy which starts with this basic philosophy:
prefer customers, then peers, then transits (of course, tier 1 carriers have no transits)
Most of them are happy to modify this policy for their customers who wish it, and usually through the use of communities, but I'm sure that it's also possible to officially request the policy change via customer service / change orders / etc - and have it set administratively (not dynamically with communities) and that's fine too.
I recommend using a combo of communities/as-prepend as the first choice, but if the customer is not skilled with this and can't get the circuit to ONLY act as a backup due to something like I described above, then simply announcing longer prefixes to the preferred carrier is an easy, guaranteed way to get the job done. (but very much a "brute force" technique that doesn't allow for fine-tuning the behavior)

I have the same query and I have tried everything mentioned on the wiki and the forums but I am hitting a roadblock. Only one of the links stays active because default routes received from both ISPs have the same distance and only one of them stays active. I have 3 clients running an identical setup and I'm facing the same issue with 2 of them. The third one surprisingly is working as i expect it to. Both router interfaces are reachable from the internet, both BGP peers are established and using PCC, I have balanced traffic between both ISPs. All 3 sites are using CCR1016-12S-1S+ with RoS 6.37.1. The routing filters are also identical on all 3 sites (allow own blocks and discard everything else in outgoing and discard own blocks and allow everything else in incoming filters) I am at my wits end as to how to resolve this. I have checked both ISP links with BGP individually on both problematic sites using ping and looking glass and the links are working fine on their own. It's just the multi-homing that's an issue.

ZeroByte I am unfortunately in a situation with my transit providers that fits what you described about the local pref.

Prepending does not work unfortunately because of the ISP upstream of our transit provider has set a high local pref for our backup transit provider over our primary where we have cheaper bandwidth.

Unfortunately I don't think the trick of announcing a /22 will work as we only have one /23 and the others are /24s that are not continuous.

I have tried making cases with both my transit providers to get the local pref changed upstream but both of them are looking like they can help me so far as it is not within their network.

Right now I am thinking the easiest way to make this work might be to just have a script that would enable filters if a peer goes down?
However I am not very good at scripting and not sure how to write it.