[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RRG] Consequences of no renumbering...
On Fri, Sep 12, 2008 at 01:54:29AM -0400, Tony Li wrote:
> |Sure. I think we're losing sight of the timetables with which we need
> |to work. I don't see a short term problem. In fact we could expend
> |resources to get us into a good long term steady state. There are
> |obvious limits on how much resources, but the argument to the business
> |manager is this: eat some additional memory here and there for now and
> |we'll see a tail-off effect (hopefully not too much) later. The key,
> |however, is completing transition.
> When, exactly, would that be? Right after the v6 transition completes? ;-)
> I suspect that the answer is 'never', in which case LISP is of zero benefit.
There may be significant gains for some sites. Suppose,
hypothetically, that all of the sites my site normally exchanges
business critical data with are all using LISP (or the solution
d'jour). If that were the case, I would seriously consider leaving the
BGP4 DFZ, and would withdraw my PI routes from the DFZ, modulo my
direct upstreams. That's a few less routes of PI space to carry about
for everyone. Lather, rinse, repeat for other PI sites, and you may
start to see some gain on a global scale.
The exercise of figuring out whether or not all of your business
critical partners are using the same solution as you is non-trivial,
but certainly possible with enough effort.
For other destinations, a default to one of your providers is probably
sufficient. My site doesn't exchange many packets with somewhereistan,
and those that are exchanged have zero need to be optimally routed.
Having said that, I'm still pessimistic about the reception much of
this work will get in the operator community. As Eliot notes, we may
need to utilize *more* resources in order to get to a world of better
scaling and fewer resources. This isn't the most altruistic group of
people on the planet (myself included). We need to have a clear idea
of why expending additional resources up front is necessary.
In a similar vein to the post about BGP transition - there might still
be a BGP3 speaker out there somewhere. I'm quite certain that I don't
care. The tail can be very long, but the vast majority of sites won't
care once the remaining volume in the tail is small enough.
to unsubscribe send a message to firstname.lastname@example.org with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg