On Fri, Nov 14, 2008 at 9:34 AM, Iljitsch van Beijnum
<[EMAIL PROTECTED]> wrote:
> Well, that was the rough consensus at the routing&addressing workshop two
> years ago. I was the only one who didn't agree. With a loc/id split
> multihoming and renumbering aversity (assuming we can contain it to the IDs
> and still renumber locs) would no longer have to increase routing tables.
> But the current routing table is 8 x the number of ASes and BGP not only
> fails to contain updates to a limited part of the network, it actually
> amplifies them as they circle the globe. Those issues are separate from an
> id/loc overload.

Hi Iljitsch,

That's an interesting point. Dissociating the GUID and SID from the
LOC would inherently suppress edge routing updates, even for basic
multihoming. However, updates due to link changes in the core still
have to propagate through the entire core, even if the change is so
distant from the instant router and the topology downstream from the
instant router is such that the change can't impact the downstream
FIBs.

I haven't heard a potentially successful solution strategy that
addresses this point. I think geoaggregation was the only one that
attempted it and we proved that geoaggregation has uncorrectable
theft-of-service anomalies even in small configurations.

The current solution strategies all imply the remaining impact from
this factor is small enough to be a non-problem. I think that's
probably right, but we should find a way to acknowledge the issue and
state the assumption explicitly.


>> changing a server's identity requires changes in many other hosts'
>> references which are unwieldy or impractical to make.
>
> Disagree: if you can run both the old and new side by side for some time,
> there's nothing to it.

I mean with the current deployed protocols. I'm trying to explain why
the routing table size problem is an emergent property of the current
protocol design.


> What I'm missing in your summary is the difference between the ID->loc
> mapping and the loc->reachability mapping. If we want to support
> multihoming, which we do or we still have a use case that can overinflate
> the routing table, we need to be able to determine which locators are
> reachable and which aren't.

Strategies A and B both depend on aggregated LOC reachability
information being flooded. It appears to be needed to compute a
network path.

I guess this ties in with your point about suppressing distant routes.
If you want to suppress distant routes then you can't flood LOC
reachability info. Is that correct?



Here's how I propose to include this knowledge in the document:

Strategy A.

Variants include:

A1a. Each core ISP has one RLOC. The RLOC's existence and reachability
is flood-propagated to the rest of the core.

A1b. Each core ISP has a small number of RLOCs for traffic engineering
(TE). The RLOCs' existance and reachability is flood-propagated to the
rest of the core.

A1c. Each core ISP has one aggregated set of RLOCs which it may
hierarchically assign to customers downstream and/or disaggregate for
TE. The aggregated RLOC's existance and reachability is
flood-propagated to the rest of the core.


Strategy B.

Assign heirarchically aggregatable LOCs to every host. Assign multiple
LOCs to each host such that in the network topography hosts appear as
stubs in multiple locations instead of forming distant connections in
the graph. Assign one aggregated set of LOCs to each core ISP where a
core ISP is one which has numerous major transit or peering links.
Flood-propagate the aggregated LOC's existance and reachability to the
rest of the core.

Having reduced the network topology to something relatively close to a
hierarchy, perform plain old hierarchical aggregation on the LOCs. Add
and remove LOCs to each host dynamically during operation as needed to
reflect changes in the nearby network hierarchies.


Strategy C.

Suppress distant routes by aggregating them into sets expected to be
available in a given direction. Because LOC reachability info is not
flooded, the routing tables each router must deal with are relatively
small.

Appears intractable.

Variants include:

C1. Geographic aggregation. All nodes within some geographic boundary
are assigned the same LOC. Routers move packets to any adjacent router
deemed to be "closer" to the LOC in question.

Geoag has been shown to have uncorrectable theft-of-service anomalies
in networks as small as 8 autonomous systems and two geographic areas.



Addendum:

Item 1. Additional causative factors worth mentioning:

Because even a distant route may present a better priority in one
coreward direction than another, it's not possible to suppress distant
routes. As a result, every default-free router in the core must carry
all routes.



Acceptable?

Regards,
Bill Herrin



-- 
William D. Herrin ................ [EMAIL PROTECTED]  [EMAIL PROTECTED]
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004
_______________________________________________
rrg mailing list
[email protected]
https://www.irtf.org/mailman/listinfo/rrg

Reply via email to