steve at ipv6canada.com
Mon Jun 7 23:07:22 UTC 2010
On 2010.06.07 18:48, Murphy, Jay, DOH wrote:
> We are obviously interpreting this in different slants.
> Definition of Transit service: for example, AS200 is said to receive transit service from, let's say AS3356, if through this connection, AS200 receives connectivity to the entire Internet and not only AS3356 and its customers.
Yes. The OP has transit through two separate ISPs. Neither of which
provide him a BGP session, because one of the providers doesn't seem
willing/capable to do so, even though the ISP who is responsible for the
space has provided the other with an LOA to allow the prefix to
originate from their ASN.
Essentially, the OP is transiting through both ISPs, but not providing
any transit services, and the transit path is provided via static routes
as opposed to dynamic ones.
> Yes I understand the customer is using static, however, some providers use BGP, and they use BGP to peer with other ISPs,
...and not only for peering, but for transit (to the DFZ) as well.
> that's it.
I have had a couple discussions with people off list. Although I don't
know the reasoning for the OP's ISP's decision to not use BGP, in cases
that I've dealt with this, it is usually due to lack of clue on how to
use private ASs, or BGP in general. These ISPs (in my experience) have
their DFZ-facing sessions set up by their upstreams, and don't have the
knowledge to configure BGP toward the clients.
Personally, if this is the case, then I'd be just as concerned with
their ability to ensure that a proper configuration to auto-detect
failure that causes removal of the prefix from their tables to avoid
blackholes. With that said, I'd also be just as concerned with their BGP
troubleshooting and filtering abilities if they were to offer a session.
Some of the smaller ISPs that fit this bill will actually allow you to
work with them and provide them advice along the way, if not even
contract the client as a consultant to ensure that this new-to-them
setup is documented properly so it can be re-used with other clients.
Also, I'm sure that it would be more work to co-ordinate the efforts for
a static setup like this between two providers than it would be to just
set up BGP. More documentation (and unnecessary static routes too).
More information about the NANOG