rproxy (Re: WOW! Re: Rsync on servers)
On Sun, Nov 11, 2001 at 03:13:22PM -0600, Adam Heath wrote:
> On 11 Nov 2001, Goswin Brederlow wrote:
> > [snip]
> Where is this client? Offloading the cpu usage to the client is a huge win.
> Also, not requiring a special daemon on the server is another plus.
Maintainer: Anand Kumria <firstname.lastname@example.org>
Depends: libc6 (>= 2.1.97)
Description: A cache which uses differences to speed up retrievals
rproxy stores hashed values of retrieved pages and when you next
access the same page it computes the difference between the
current page and the recently stored page.
The rproxy extensions to HTTP allow the server to generate a hsync
relative to the cached instance in a way that is completely general,
and transparent to both the server and user agent.
rproxy, and clients and servers which implement hsync, calculate a
block-by-block signature of the file, by computing a checksum over
consecutive extents of equal length, such as 1024 bytes. This checksum
is then added into a header of the request and transmited as usual.
To be useful, there should be at least two rproxy instances between
the client and the server. Transfers between the proxies will be
delta-encoded, while the browser and server will just see standard HTTP.
For example, it is very useful to run on instance on each side of a modem
link, so that data across the slow link will be delta-encoded. Further
information is available at http://linuxcare.com.au/rproxy/
Note that the checksums are transmitted along with the request, so they are
first computed by the client.