Browse Prior Art Database

Heuristic (Intelligent) Operating System Control Technique of Start/Stop I/O

IP.com Disclosure Number: IPCOM000040727D
Original Publication Date: 1987-Dec-01
Included in the Prior Art Database: 2005-Feb-02
Document File: 3 page(s) / 36K

Publishing Venue

IBM

Related People

Batalden, GD: AUTHOR [+4]

Abstract

This technique is used to dynamically control terminal response time in connection with a data processor. The goal is, of course, to first minimize the terminal response time. However, as the workload increases, less processor time is available to service workstation I/O. Therefore, as workload increases, it is desirable to slow terminal response time in a controlled manner instead of getting into a situation where there is thrashing and response time becomes very erratic, unpredictable, and possibly very different between workstations. The technique very efficiently monitors both the time the processor spends servicing terminal requests and the overall system workload. As the processor becomes busier, the service to terminals is reduced by reducing the baud rate which is being used.

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

Heuristic (Intelligent) Operating System Control Technique of Start/Stop I/O

This technique is used to dynamically control terminal response time in connection with a data processor. The goal is, of course, to first minimize the terminal response time. However, as the workload increases, less processor time is available to service workstation I/O. Therefore, as workload increases, it is desirable to slow terminal response time in a controlled manner instead of getting into a situation where there is thrashing and response time becomes very erratic, unpredictable, and possibly very different between workstations. The technique very efficiently monitors both the time the processor spends servicing terminal requests and the overall system workload. As the processor becomes busier, the service to terminals is reduced by reducing the baud rate which is being used. Reducing the service to terminals will effectively avoid most overcommittment situations.

(Image Omitted)

This technique was developed specifically for ASCII devices which utilize processor interrupts to move data to or from the async line. While the techniques may have some applicability to other situations, they were particularly important in this case because of the processor load required to service the many interrupts generated by the async- connected devices. The implementation of this technique is to control the baud rate used for each of the devices attached to the system via an async connection. Varying the baud rate will directly affect the processor workload, because there is essentially one interrupt to the processor for each character. The problem, of course, is to control the baud rate in a reasonable manner without adding extensive overhead.

(Image Omitted)

In addition to the direct reduction in processor load, there are secondary effects which also reduce the load. First, the work requests from the terminal users are also reduced by this action. Second, data transfer over-runs are also reduced or eliminated. Avoiding over-runs is important because of the system cost of the error recovery procedures. To facilitate the baud rate modification, the first requirement is a table listing each of the attached devices and their maximum permitted baud rate. The maximum rate can, of course, be determined by either the device or the line connecting the device. This table is constructed during system initialization from the system configuration data (see Fig. 1). To facilitate modified baud rates, the device table is expanded to include an entry for the rate currently used by the device and a minimum rate for the device. During system initialization, each of the devices is set to operate at its maximum baud rate. Of course, a cumulative rate of all devices at their maximum is not possible to support. However, initially not all devices will be operating, and it is desirable to give th...