Browse Prior Art Database

Hypervisor High Performance Synchronous Dispatch for Coupled Systems

IP.com Disclosure Number: IPCOM000118311D
Original Publication Date: 1996-Dec-01
Included in the Prior Art Database: 2005-Apr-01
Document File: 2 page(s) / 44K

Publishing Venue

IBM

Related People

Baskey, ME: AUTHOR [+6]

Abstract

Disclosed is a method for improving the performance of dispatching a Process Resource/Systems Manager (PR/SM) logical partition. The new method provides a performance benefit when one logical partition issues a synchronous request for service from another logical partition on the same processor.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 69% of the total text.

Hypervisor High Performance Synchronous Dispatch for Coupled Systems

      Disclosed is a method for improving the performance of
dispatching a Process Resource/Systems Manager (PR/SM) logical
partition.  The new method provides a performance benefit when one
logical partition issues a synchronous request for service from
another logical partition on the same processor.

      On a logically partitioned processor, one partition may be
initialized as a server which processes synchronous requests from
other partitions.  Synchronous in this case means that the requesting
partition spins (consuming CPU time) until the server partition has
processed the request and returned the response.  Thus, any delay in
getting the request to the server partition can result in wasted
processor cycles in the requesting partition, as well as reduced
throughput.

      On some processors, particularly uni-processors, only one
partition can be active at a time.  Since the processor of the
request (server) requires control to receive and process the request,
it has to wait until the requestor's time slice has expired within
the hypervisor.  As a result, the effective throughput of the system
suffers.

      To address this problem, a new dispatching mechanism called the
"directed dispatch" has been developed.  It involves changes to the
hypervisor in the area of the dispatching of partitions.

      When the hypervisor detects that a logical partition has issued
a synchr...