2002-09-05 11:46:10 +00:00
|
|
|
<title>Work distribution</title>
|
2002-08-20 23:54:17 +00:00
|
|
|
<body bgcolor=ffffff>
|
2002-09-05 11:46:10 +00:00
|
|
|
<h2>Work distribution</h2>
|
2002-04-30 22:22:54 +00:00
|
|
|
<p>
|
2002-08-19 18:43:10 +00:00
|
|
|
Results are not necessarily dispatched one at a time.
|
|
|
|
|
Instead, each participant host
|
|
|
|
|
maintains an estimate of the amount of work remaining
|
|
|
|
|
(i.e. the time until one of its processors will be idle).
|
|
|
|
|
Each host also
|
2002-07-29 19:01:38 +00:00
|
|
|
has two scheduling parameters, the <b>high-water mark</b> and the
|
2002-08-20 23:54:17 +00:00
|
|
|
<b>low-water mark</b>
|
|
|
|
|
(these are part of the host's "preferences", discussed elsewhere).
|
2002-04-30 22:22:54 +00:00
|
|
|
<p>
|
2002-08-19 18:43:10 +00:00
|
|
|
Normally the work remaining is between the high- and low-water marks.
|
|
|
|
|
When the work remaining reaches the low-water mark, the core
|
2002-08-20 23:54:17 +00:00
|
|
|
client contacts one or more scheduling servers,
|
|
|
|
|
and attempts to get enough work to exceed the high-water mark.
|
2002-08-19 18:43:10 +00:00
|
|
|
The scheduling server sends a maximum of four weeks of work.
|
|
|
|
|
If a work unit is not feasible for a host
|
2002-08-20 23:54:17 +00:00
|
|
|
because it consumes too many system resources,
|
|
|
|
|
the scheduling server will not send that work unit.
|
2002-04-30 22:22:54 +00:00
|
|
|
<p>
|
2002-08-19 18:43:10 +00:00
|
|
|
The amount of work (measured in FP/int ops) sent in a single
|
|
|
|
|
scheduling RPC can be limited by the project, on a per-application basis.
|
2002-04-30 22:22:54 +00:00
|
|
|
<p>
|
2002-08-20 23:54:17 +00:00
|
|
|
This scheme allows hosts that are sporadically connected
|
|
|
|
|
(because they're portable or have modem-based connections)
|
|
|
|
|
to avoid becoming idle due to lack of work.
|
