Redundant Web services?

Rich Wiggins (WIGGINS@msu.edu)
Mon, 06 Dec 93 14:55:26 EST


There have been various discussions of caching and redundant services,
and a lot of this dovetails with Uniform Resource Identifier issues.
I'm curious if there are any prospects for a simple scheme for
implementing redundant servers in Webspace.

The Gopher protocol provides a scheme for identifying a redundant
service for a given document. Believe it or not, this is in the
original Gopher spec -- the "+" data type (not to be confused with
Gopher+). If an item has a plus type, then it's a redundant
location for the previous document.

This seems straightforward and very useful. To my knowledge, it's
never been implemented. Is there any chance that the Web community
could define a similarly straightforward scheme, and actually get it
implemented?

The particular inspiration for this is that lately we've had a lot
of trouble connecting to the home page at CERN. A few weeks ago
there was an all-day outage of NCSA due to power failure. Mirroring
of key sites would be very useful not just in terms of outages, but
also in the interest of load-balancing.

/Rich Wiggins, CWIS Coordinator, Michigan State U