Electrical computers and digital processing systems: multicomput – Computer-to-computer data routing – Least weight routing
Reexamination Certificate
1997-09-10
2004-03-09
Follansbee, John (Department: 2126)
Electrical computers and digital processing systems: multicomput
Computer-to-computer data routing
Least weight routing
C709S226000, C709S229000
Reexamination Certificate
active
06704766
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method and apparatus for dynamically controlling the execution of a request handler on a processor resource. More particularly, the invention relates to a method and apparatus for dynamically controlling the dispatching of the polling loop of a coupling facility operating in a logical partition of a partitioned information handling system
2. Description of the Related Art
Many computer hardware machines conforming to the IBM® S/390® architecture, as described, for example, in the IBM publication
Enterprise Systems Architecture/
390
Principles of Operation
, SA22-7201-02, December 1994, incorporated herein by reference, operate in what is known as logically partitioned (LPAR) mode. Logically partitioned computer systems are well known in the art and are described in Guyette et al. U.S. Pat. No. 4,564,903, Bean et al. U.S. Pat. No. 4,843,541, and Kubala U.S. Pat. No. 5,564,040, incorporated herein by reference. Commercial embodiments of logically partitioned systems include IBM S/390 processors with the Processor Resource/Systems Manager™ (PR/SM™) feature and described, for example, in the IBM publication
Processor Resource/Systems Manager Planning Guide
, GA22-7236-01, June 1997, incorporated herein by reference.
Logical partitioning allows the establishment of a plurality of system images within a single physical machine, or central processor complex (CPC). Each system image is capable of operating as if it were a separate computer system. That is, each logical partition can be independently reset, initially loaded with an operating system that may be different for each logical partition, and operate with different software programs using different input/output (I/O) devices. Logical partitioning is in common use today because it provides its users with flexibility to change the number of logical partitions in use and the amount of physical system resources assigned to each partition, in some cases while the entire central processor complex continues to operate.
A recent addition to the IBM S/390 architecture, usually operating in a logically partitioned environment, is the IBM Parallel Sysplex™ configuration, comprising two or more systems interconnected via a coupling facility to form what is known as a “sysplex” (for “system complex”). A sysplex configuration may have more than one coupling facility, for example a backup coupling facility set to take over if a primary coupling facility fails. Each system that is a member of a sysplex may be either a separate hardware machine or a separate logical partition of a particular hardware machine. In a similar manner, each coupling facility in the sysplex may be either a separate hardware machine or a separate logical partition of a particular hardware machine. Each S/390 coupling facility, whether a separate hardware machine or a separate logical partition of a particular hardware machine, is implemented by microcode known as coupling facility control code (CFCC).
In the S/390 Parallel Sysplex architecture, a system issues a request to a coupling facility using a Send Message (SMSG) instruction. This instruction is executed by a central processor (CP) of the machine on which the system resides; this CP may be a logical CP if the system resides in a logical partition. The executing CP causes a message command block (MCB) to be sent along a message path to the coupling facility and receives back a message response block (MRB) containing the results of the request. Message communication and other aspects of the operation of an S/390 coupling facility are described in such references as Elko et al. U.S. Pat. No. 5,561,80, Elko et al. application Ser. No. 08/147,697, filed Nov. 4, 1993, and the patents and applications referred to therein, all incorporated herein by reference.
SMSG instructions may be executed either synchronously or asynchronously. When executed asynchronously, the SMSG instruction is competed as soon as the request in the form of an MCB is sent off to the target coupling facility. When executed synchronously, on the other hand, the SMSG instruction is not completed until a response in the form of an MRB is received back from the target facility. The handling of synchronous and asynchronous requests in an S/390 Parallel Sysplex environment is described further in the copending application of applicant J. P. Kubala et al., Ser. No. 08/903,285, filed Jul. 30, 1997, incorporated herein by reference.
The responsiveness of a coupling facility is a critical factor in the performance of a S/390 Parallel Sysplex configuration. The majority of coupling facility (SMSG) requests from an originating system are synchronous in nature, that is, the issuing processor remains in a busy state while waiting for the coupling facility request to complete. Thus, the coupling facility response time has a direct impact on the system processing cost of coupling facility requests.
Coupling facility response time has two components: (1) the time elapsed before the coupling facility control code recognizes that a request has arrived and is ready to be processed; (2) the time the control code spends processing the request. In order to minimize the first component (recognition time), the control code constantly runs a “polling loop” on each processing engine of the coupling facility to look for newly arrived requests. While effective in minimizing recognition time, this polling loop can consume all processor resource allocated to a coupling facility partition.
For high availability, multiple coupling facilities are required in a Parallel Sysplex configuration. In the event of the failure of one coupling facility, there must be one or more alternative coupling facilities to take over the load of the failed coupling facility. Often a cost-effective configuration will include primary and backup coupling facilities, where the backup coupling facility is asked to process coupling facility requests only in the event of the failure of the primary coupling facility. Furthermore, the backup coupling facility can be defined as a partition with one or more shared processors on the same central processor complex with system partitions (i.e., those partitions containing an operating system such as OS/390 rather than coupling facility control code).
Providing a backup coupling facility partition on the same machine as the system partitions creates a problem, however. If sufficient processor resources are made available to the backup coupling facility partition so that it can process the full load in the event of a failure of the primary coupling facility, then the control code polling loop will consume all those resources the rest of the time, making them unavailable to the system partitions. On the other hand, if only a small amount of processor resource is made available to the backup coupling facility partition, it cannot be responsive to a primary coupling facility failure condition. A similar problem arises when defining coupling facility partitions that only need to process very low coupling facility request rates (for example, for testing purposes).
SUMMARY OF THE INVENTION
In general, the present invention contemplates a method and apparatus of dynamically controlling the execution of a request handler on a processor resource in an information handling system in which one or more requesters issue requests to a request handler executing on the processor resource. In accordance with the invention, a plurality of request handler execution modes are defined having differing amounts of utilization of the processor resource. Initially, one of the execution modes is selected, and the number of requests received from the requesters within a predetermined interval while executing the request handler in the selected mode is determined. Transitions are made between the request handler execution modes in accordance with the number of requests received within the predetermined interval in the selected mode.
The requesters and the request handler may reside in different logical part
Goss Steven N.
King Gary M.
Kubala Jeffrey P.
Nagy John C.
Surman David H.
Bullock, Jr. Lewis A.
Follansbee John
Kinnaman, Jr. William A.
LandOfFree
Method and apparatus for dynamically controlling the... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for dynamically controlling the..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for dynamically controlling the... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3240747