Browse Prior Art Database

Workstation Load Leveling Technique Using Buffer Allocation

IP.com Disclosure Number: IPCOM000040725D
Original Publication Date: 1987-Dec-01
Included in the Prior Art Database: 2005-Feb-02
Document File: 3 page(s) / 57K

Publishing Venue

IBM

Related People

Batalden, GD: AUTHOR [+4]

Abstract

This technique helps to more effectively utilize a data processor by avoiding unproductive queueing. When work is arriving at a rate which can be handled effectively, work queues remain small. However, when work is arriving at a faster rate, the new work must be added to already long queues. The time spent queueing further reduces the time available for processing. The technique described here helps to eliminate the unproductive queueing time - therefore increasing effective throughput. As a secondary effect, the number of interrupts is also reduced which further increases the throughput. At a high level, the technique is to artificially limit the number of workstation requests which will be accepted. After a fixed minimum number of buffers are in use, acceptance of additional requests is limited.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Workstation Load Leveling Technique Using Buffer Allocation

This technique helps to more effectively utilize a data processor by avoiding unproductive queueing. When work is arriving at a rate which can be handled effectively, work queues remain small. However, when work is arriving at a faster rate, the new work must be added to already long queues. The time spent queueing further reduces the time available for processing. The technique described here helps to eliminate the unproductive queueing time - therefore increasing effective throughput. As a secondary effect, the number of interrupts is also reduced which further increases the throughput. At a high level, the technique is to artificially limit the number of workstation requests which will be accepted. After a fixed minimum number of buffers are in use, acceptance of additional requests is limited. Specifically, additional requests are accepted only when processor utilization indicates that more input requests can reasonably be handled. This contrasts with a "normal" buffer management scheme which would somehow insure that all required buffers for a "worst-case" situation would be available.

(Image Omitted)

In the environment where this technique may be applied, one processor is creating work requests and then passing them on to a second processor acting as a workstation controller for further action. The first processor is processing a variety of jobs, and only communication of workstation requests is blocked while waiting for the second processor to respond to a work request. Workstation requests are received by the second processor as an interrupt. They are immediately queued, and later handled on the main processing level. This second processor also has other I/O responsibilities however, so it is well utilized without handling any workstation requests. The second processor has of course finite capabilities. It would be theoretically possible for requests to arrive so fast that this processor would be busy 100% of the time simply queueing up the workstation requests - without actually processing any request. Of course, this situation would also rapidly fill the available memory. It is also necessary to understand the technique to note that this processor is operating in an environment which includes a storage manag...