[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [GROW] ORTC



Thanks for the pointer to the paper! I just had a quick read. The ORTC algorithm passes the tree 3 times and gives the optimal aggregation when no extra routable space is allowed. The table size after aggregation is around 60-70%.

Our level-4 result, however, shows that if extra space is allowed, the table size after aggregation is 30-40%.

It'll be great if you could send us the ORTC code. We're very interested in comparing it with other schemes under the same environment.

Thanks,

---
beichuan

On Nov 11, 2009, at 11:23 AM, Beichuan Zhang wrote:

Quick synopsis:

ORTC performs FIB compression such that the number of entries in the compressed FIB is the smallest. However, ORTC requires 3 passes if a tree data structure is used to store the prefixes.

The compression achieved is roughly of the order of what is achieved by level-4b (about 70% on recent tables from routeviews) but no "new" routable space created that was previously unroutable.

On Wed, Nov 11, 2009 at 7:02 AM, Zartash Uzmi <zartash at gmail.com> wrote:
I have a working implementation of ORTC and can make it available if needed. The implementation has been tested for routing correctness, on a large number of routeviews tables.

Zartash


On Wed, Nov 11, 2009 at 6:52 AM, John G. Scudder <jgs at juniper.net> wrote:
Constructing Optimal IP Routing Tables
Draves, R.P.   King, C.   Venkatachary, S.   Zill, B.D.
INFOCOM '99

http://research.microsoft.com/apps/pubs/default.aspx?id=69698

(I got the acronym expansion wrong, it's "Optimal Routing Table Constructor".)

--John