Two ISP lines
Randall R. Stewart (home)
randall at stewart.chicago.il.us
Mon Jan 26 05:57:54 PST 2004
Juan Rodriguez Hervella wrote:
>Just some questions about this paper:
>"(...) Note that this route has to be at the same level of the tree, i.e.
>the code cannot return a less specific match or a more specific
>Question: I don't understand why if you are looking for an alternate
>route you aren't allowed to retrieve a more specific route. This doesn't
>make sense to me. If you are routing packets using a route when
>there is a more specific match, you aren't doing "longest prefix match".\
When you first ask for a route.. you get the longest prefix
match. So you are at say a depth in the tree of N. If you
try to go to a lower level in the tree (say N-1) then you
could be going to the wrong place... I suppose one could
go to a N+ level in the tree.. i.e. more specific match (assuming
that maybe a new route was added lower in the tree).. but I
never contemplated that aspect much because it would
be more of a corner case I think... It would also make
the code a bit more complex as well :->
>Another question: if ISP-1 goes down, and you use this feature of
>alternatives routes, this still doesn't fix the communication problem. Unless
>you make something with the source addr. of the multihomed site's
>packets, the reply packets will be lost in the faulty ISP, imho.
Ahh.. it is true that SOME ISP's do ingress filtering.. for TCP
this would be a problem.. But for SCTP (a multi-homed protcol)
this is NOT a problem.. the SCTP association would just
use the source address of the outbound interface (since the
association is made up of a SET of addresses on each side
and a port on each side)....
Now there are still ISP's that do NOT filter (I have one that
does and one that does not :->)...
>On Friday 23 January 2004 01:38, Randall R. Stewart (home) wrote:
>>An interesting question... the following link has
>>some thoughts along these lines... and something
>>for the BSD community to think upon...
>>TCP could definetly use something like the above (with Itojun's Multi-path
>>updates as well).. it would give more reliability to even a singly
>>homed protocol such as TCP :->
>>Andrea Venturoli wrote:
>>>Ok, I asked already asked something similar to this in the past, but it's
>>>not the same thing... maybe it's a trivial question...
>>>If I had two lines to the Internet: how would I use both?
>>>Could I just provide two default routes? How?
>>>What algorithm would be used to choose among the two?
>>>What if one failed?
>>>bye & Thanks
>>>freebsd-net at freebsd.org mailing list
>>>To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
Randall R. Stewart
815-342-5222 (cell phone)
More information about the freebsd-net