Here are two of
the reasons why I think RO will be undesirable:
1. It allows the
users to entirely bypass the home IP provider's network. This will keep the home
IP network providers out of added revenue streams. The situation will be even
worse when the AAA clients for accounting are in the home IP network which will
not be in the data path due to RO.
2. Imposes
unnecessary processing requirement on ALL IPv6 devices to support this
non-mandatory functionality.
Here is
a reason why I think RO will be meaningless:
The core of the
internet is managed by large carriers. These carriers use (or will be
using) Constrained Based Routing (OSPF-TE) instead of plain OSPF. The
main purpose is traffic engineering. Therefore the path between the CN and
the MN may not always be the shortest one even with RO. It entirely depends on
the traffic engineering of the intervening networks. As an
example:
If the CN is in
Chicago, MN is in Dallas and the HA is in Miami, if the shortest route between
Chicago and Dallas has less weight (not preferred by OSPF-TE) then all the IP
packets between CN and the MN will be routed via an alternative path which may
well be via Miami. Therefore RO will be a waste of time and
resource.
Regards,
Kuntal
>Michael Thomas wrote:
>> Frankly, I don't think that there is any evidence
>> that the net would be substantially harmed if RO
>> wasn't widely implemented and/or enabled. Indeed,
>> I think there's good reason to believe that many/most
>> nodes will not enable RO even if their kernel
>> implements it. In some cases, it's likely to be
>> a nice and useful optimization, but I really
>> don't see it as a "if we don't do this the net
>> will fall apart". As such, SHOULD seems like it
>> strikes the right balance.
>>
>> Mike
>>
>
>I could not agree more with this point. RO would be a required
>functionality a number of years back, when the net was in it's
>infancy and bandwidth availability in the backbone networks
>(even transoceanic links) were severely limited.
>We are in the age of OC-192/768s over DWDM. The delay that an IP
>packet incurs in the core is almost negligible.
>If bottleneck in the HA is the primary driver for RO, then it is
>possible to solve the problem with load balancing with DHAAD.
>
>My two cents.RO gives you the capability of having two end-points connected without
having to rely on intermediate nodes such as the HA being
involved. The HA is not the bottleneck. One of the benefits of RO is
lesser traffic in the backbone and to a certain degree reduced
latency. But these are some of the lesser motivations.Having a mobile anchored at some point in the network (HA) when it is
non-essential is just not the right approach. Mandating HAO processing
in all IPv6 nodes is the only way to accomplish this and reverse
tunnelling or having a node anchored at the HA is an option only for
backward compatibility to the nodes that already deployed (which is a
pretty small percentage of IP nodes).>
>Regards,
>Kuntal
>-Basavaraj
