how hint caching works

It occurs to me that I could at least explain how the
speedup of Hint Cache vs. ICP or Cache Digest works. Sorry, I wrote
letters about pushcache to both non-programmers and to you guys that
same day; I sometimes forget to put to put the details in the
letters to people who appreciate it, and to keep it out of letters to
people who don't. Irritates everybody.

A hint cache knows where the closest copy of every single object in a
cloud of hint caches is.

That allows cross-cache access to be very efficient. If an object is
not in the cache, it looks in its hint cache. If any other cache in
the cloud has a copy, it goes to the closest place. Otherwise, it
goes directly to the original server, without passing GO or going
through siblings or parents or kids.

Once the cache has its own copy, it advertises that fact.

Unlike cache digests, hints are propagated through the entire cloud.
(we have done simulations that show that this is surprisingly scalable).
For more details, go to www.pushcache.com/wp.html and read up on the
paper by Renu Tewari&co.

Unlike cache digests, hints have a notion of version through
last-modified time, and always go for the most recent known version.
They also have some other handy metadata.

The only way you get better pull performance than using hint cache is,
well, by *push*ing things in advance. So this is an architecture
suited to speeding things up for the future. People have actually
used pushcaches to speed up pull rates.

-- 
Jon Kay        pushcache.com                      jkay@pushcache.com
http://www.pushcache.com/                             (512) 420-9025
Squid consulting				  'push done right.'