Default Routes with full BGP Routing table

[nygenxny123]

We currently have  2 ISP providers giving us a full routing table for our data center internet access

However, we just upgraded one internet link

Now we have

300Mb-ISP-A

1Gig-ISP-B

After the upgrade I noticed that ISP-A was sending us a default route along with the full BGP routing table

ISB-B was not sending us a default route along w/ the BGP table.

Router B was learning the default route via router A via BGP/OSPF relationship betweek our two routers

ISP-A also appears to be sending us more routes

My question is...Would it be beneficial to have ISB-B also send us a default route?

[ghostinthenet]

If both providers are giving you a full BGP feed, there's no advantage to receiving a default route from either. Anything that isn't in either feed is going to be unreachable, so a default route doesn't gain you anything. Without a default route in this scenario, your edge routers can report a network unreachable error to your clients, which will allow for faster troubleshooting when it comes to reachability issues.

Personally, I would either ask the provider that is sending you the default route to stop, or filter it inbound. I would also be looking into why ISB B isn't sending you as many prefixes.

[Giuseppe Larosa]

Hello Jody,
many years ago I have used the fact of receiving a default route in BGP in addition to the BGP full tables in order to conditionally generate an O E1 default route in OSPF with a different seed metric depending on what the BGP next-hop of the 0.0.0.0/0 eBGP route was. So a possible use for the BGP defaul route may exist in special scenarios. I agree that or both ISPs do send it or both do not. And yes for troubleshooting purposes could be better not to have a default route

About full BGP internet tables : each provider sends its own view of the BIG internet.
Different ISPs provide different number of Ipv4 prefixes, if the difference are small in percentage for example 650,000 routes against 660,000 routes it is normal.

IF the number of routes received by ISP B is much less then the ones sent from ISP A:
as Paul has noted there is a possibility you have become a transit AS for ISP B, but you should see your link utilization to ISP B full.
A second possible case is that ISP B has actually a relationship with ISP A and most of its routes are learned by ISP A.

Ideally, the best results are achieved if ISP A and ISP B have the same "size" in the internet, if ISP A is a Tier 1 and ISP B is a Tier2, ISP A AS paths will be shorter in average specially if ISP B sees IPS A as one of its own upstream provider.

Best Regards
Giuseppe Larosa

[paul driver]

Hello

It is possible that one of your ISP's are now using your rtr and ASN as a transit path , You need to make sure your only adverting your local originated routes to either ISP and not any ISP routes also.

example:
ip as-path access-list 10 permit ^$

route-map Local_Routes-Out permit 10
match as-path 10

router bgp xx
neigbour x.x.x.x. remote-as (isp1) route-map Local_Routes-Out out
neigbour x.x.x.x. remote-as (isp2) route-map Local_Routes-Out out

or 
neigbour x.x.x.x. remote-as (isp1) filter-list 10 out
neigbour x.x.x.x. remote-as (isp2) filter-list 10 out

res

Paul

[ghostinthenet]

Even simpler:

route-map RM_Filter_In permit 10
set community no-export
!
router bgp xx
neigbour x.x.x.x route-map RM_Filter_In in ! ISP1
neigbour x.x.x.x route-map RM_Filter_In in ! ISP2

[Joseph W. Doherty]

As the other posters have noted, I too would desire both ISPs be consistent.

Further, as noted in one of the other posts, having a default, provided by BGP might make it a bit easier to inject a default into your IGP.

Lastly, as you note you have 300 Mbps and gig ISP connections, I would suggest you look into PfR. If you do pursue and use PfR, you no longer need anything but a default route from your ISPs.