Re: We're putting lots of work into caching, how about the server side?

At 12:24 AM 17/12/96 +0000, Miguel A.L. Paraz wrote:

>P.S. I'm also concerned about the rising popularity of "push" technologies
>such as Pointcast. Would the vendors be interested in putting caching
>in these protocols?

  Actually, there are classes of information, such as news in its non-Usenet
sense, which it is entirely appropriate to distribute in this way. Well, via
multicast anyway. I suspect the best answer is not to try getting caching
put into a multicast protocol, but to have the multicast protocol used where
appropriate to distribute URN's, which may then be retrieved through
orthodox HTTP, with caching taking its place(s) in that transfer.
  That would work well as a replacement for Usenet, and probably also for
applications like much of the content distribution done by organisations
like Reuters.
  But when it comes to sending out really important material like the
forthcoming defeats of the England cricket team on tour, multicast,
generally without caching, is the right answer.

- Donald Neal

--
Donald Neal               | Anita and Donald Neal are delighted to announce
Systems Programmer        | the birth, on Friday, 27 September, of Peter
The University of Waikato,| Alexander Neal (3470g), younger brother of
Hamilton, New Zealand     | Martin. Mother and baby are both very well, if
                          | not sleeping as much as you might hope.
---------------------------------------------------------------------------