[c-nsp] Inbound redundancy with two ISPs

[The Father]

Garry wrote:
> The Father wrote:
>> Is multihoming a valid reason even if they can't justify a /24 worth 
>> of IP addresses?  I would have thought that ASNs were hard to get 
>> since there's a finite number of them (currently anyways).
>>   
> Well, most likely they will continue to be finite, even if they are 
> migrated/extended to 32bit ... ;) FWIW, 1 AS per 1.5 people on earth 
> is pretty close to infinite *gg*
>> Oh and I forgot to mention in my original post that is there still a 
>> valid solution if BGP is not an option?
>>   
> Not an option as in "full table" or not an option period?
>
> The main problem with not working w/ BGP will probably be reliably 
> stopping announcements of your IPs from the ISP whose link to you is 
> down ... there are ways to get around it, but using BGP for at least 
> receiving a default route and announcing the prefix would definitely 
> make the whole issue easier to handle and implement ...
>
> As for outgoing traffic, you could also use tracking to redirect 
> outgoing traffic - do icmp checks of the next hop, and configure the 
> default routes with the appropriate tracking object ...
>
> -gg
>
> .
>
Not an option period.  I'd look at it for example, as the customer 
having a T1 connection from the other provider and they're using a /28 
for all their servers.  Their ISP may not qualify them for regular 
public BGP.