From: Slavik Krassovsky (viatk_at_windows.microsoft.com)
Date: Mon Mar 01 2004 - 17:23:50 PST
In this 1997 paper Bugnion et al discuss a possible solution to the
problem of huge costs of development of operating systems for new (and
especially shared memory massively multiprocessor) hardware. A new layer
or level of indirection is introduced that serves as a multiplexer of
system's resources to a collection of commodity operating systems that
run on the machine all at the same time.
The paper is very good - it's very clear to see the motivation of the
project and the accomplishments. The problems I encountered - my
unfamiliarity with NUMA architecture - will have to look it up, the
convoluted test bed - unfortunately the FLASH machine was not available
to the authors and they used some other machine - immediately it created
a suspicion that the results "may vary" once a real machine is used.
This archive was generated by hypermail 2.1.6 : Mon Mar 01 2004 - 17:23:57 PST