On Jan 28, 2009, Dino Farinacci wrote:

You cannot push around 10^10 entries and store them everywhere. [...]



Dino -

I think we should experimentally compare ALT with other mapping systems
before we decide whether pull-based or push-based mapping systems are
better.  Dismissing push-based mapping systems without corroborating
data would be a bit premature in my opinion.

In the absence of experimentation results, I would actually argue in
favor of push-based mapping systems based on some analytical reasoning:
Pull-based mapping systems have two important disadvantages compared to
push-based mapping systems:

- Performance penalty, i.e., additional propagation latencies for some
  packets, and higher loss probabilities for packets that take a sub-
  optimal path

- Robustness penalty due to a new online dependency on components off
  the actual transmission path.  (FWIW: All pull-based mapping systems
  have this penalty.  Mapping requests must be routed via overlay
  infrastructure because the direct route is unknown at that time.)

Furthermore, I do not share your concerns regarding push-based mapping
systems:  BGP is pushing routing data already today, and this works
fine.  Any routing-scalability-related issues with BGP are not due to
BGP being push-based; they are due to frequent updates and a high load
for core routers.  Both of these issues would go away in an address
indirection architecture (be it LISP, Ivip, APT, or Six/One Router),
independent of whether you use a pull- or push-based mapping system.

In conclusion:  The overlay approach of ALT is certainly an interesting
idea.  But I think it would be premature to conclude that it is the only
viable solution before we have more evidence to back this claim.

- Christian


PS:  I admit that I have never been really good in avoiding cross-
posting.  But this is obviously my all-time negative record...



_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to