RE: Scalability issues in the Internet routing system - #2

Andre:

#2 FIBs doesn't have to be impacted by Moore's law.

In certain times of routing technology people "cooked" routing tables in
order to solve #2. The key there is to run through the proposed FIB
table
Generated by a routing table and cook it for FIB insertion. You can
often
summarize (from the viewpoint of a particular router) all the routes
into
a manageable set of FIB entries.

The issue is how much fluctuation. 10% (actually 8-14% but 10% is
average) of the Internet (see route-views, etc). FIB hardware can
allow you to split
the prefix load and the next-hops. So, you load prefixes that fluctuate
and
only break the prefix-next-hop when they fluctuate.

.. bottom line.. I don't think we will be limited by Moore's law for
#2..
If we utilize existing technology existing in Hw and software
methodology,
we can change the game on #2.

Sue Hares

PS - My answers tend to be brief - glad to expand in private or public.
     Your choice.

Susan Hares wrote:

PS - My answers tend to be brief - glad to expand in private or public.
     Your choice.

Please expand here on the list. That's very on-topic for NANOG actually.