Electrical computers and digital processing systems: memory – Storage accessing and control – Memory configuring
Reexamination Certificate
2002-06-06
2004-06-01
Ellis, Kevin L. (Department: 2188)
Electrical computers and digital processing systems: memory
Storage accessing and control
Memory configuring
Reexamination Certificate
active
06745312
ABSTRACT:
COPYRIGHT NOTICE
A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all copyright rights whatsoever.
BACKGROUND OF THE INVENTION
The invention disclosed herein relates generally to methods for managing resources within a computer system. More particularly, the present invention relates to a system and method for managing resources within a computer system having multiple partitions.
Through logical partitioning, a physical computer system may be divided into two or more logical partitions (“LPARs”), each of which has a share of the physical system's resources and operates as an independent logical computer system within the physical system, having also, for example, its own operating system, user applications, data files, primary language, time of day, etc.
The share of resources allocated to a partition affects the performance of the logical system within that partition, and a lack of one or more resources, such as processor power or memory, may result in performance degradation. Also, since work loads, and consequently, resource needs, may change over time, one defined allocation of resources may not be effective for all situations. Thus, it is advantageous to be able to reallocate resources among partitions to address changes in work loads.
Currently available partitionable systems, such as the AS/400™ systems running the OS/400™ operating system, both from International Business Machines Corporation, support the dynamic reallocation of resources among partitions, e.g., without having to perform an initial program load (“IPL”) afterwards, but do not provide sophisticated tools to execute such reallocations. Generally, dynamic reallocation of resources is performed either at the discretion of a human operator or automatically according to a schedule, e.g., automatically executing a preset resource reallocation based on assumed needs at predetermined times. The first technique is deficient in that it requires a human operator and is time consuming. The second technique, although not requiring an operator after the preset resource reallocations have been defined, has the drawback of reallocating resources based on assumed needs and not on current actual needs of the partitions.
Thus, there is a need for a system and method that automatically determines when partitions suffer performance degradation due to a lack of resources, and dynamically reallocates resources among the partitions to alleviate the identified performance degradation.
BRIEF SUMMARY OF THE INVENTION
It is an object of the present invention to provide a system and method that dynamically reallocates resources among partitions in a way that avoids the drawbacks described above.
It is another object of the present invention to provide a system and method that automatically determines when partitions suffer performance degradation due to a lack of resources and provides an indication of the level of need for resources.
It is another object of the present invention to provide a system and method that automatically and dynamically reallocates resources among partitions according to an indicated need for resources.
The above and other objects are achieved by a system and method for automatically reallocating resources (“tuning”) among partitions based on factors including need for resources (e.g., as indicated by resource pain rates) and user assigned priority. The system includes a tuner, a storage, and partition interfaces for one or more partitions to be tuned. The partition interfaces continually obtain and store in the storage data from the partitions to be tuned regarding their performance with regard to resources, including memory, processor power, and interactive power. The tuner executes several tasks, including a validity table creation task, a memory tuning task, and a processor power and interactive power (“P/I”) tuning task.
In the validity table creation task, for each allocatable amount of processor power, the tuner determines the minimum and maximum amounts of interactive power, if any, that can be validly allocated in combination with the respective allocatable amount of processor power. Then, for each allocatable amount of processor power for which can be determined both minimum and maximum amounts of interactive power that can be validly allocated in combination with the respective allocatable amount of processor power, the tuner adds an entry in a table, the entry including the respective allocatable amount of processor power and the minimum and maximum amounts of interactive power determined for the respective allocatable amount of processor power.
In the validity table creation task, the tuner creates a validity table useful for the P/I tuning task or which may be used by itself to allow someone, e.g., a system administrator, to manually tune processor power and interactive power across partitions. The validity table created through the validity table creation task shows, for a given system configuration, the amounts of processor power and interactive power that may be validly allocated in combination to a partition. The tuner executes the validity tables creation task to create a validity table at the time the tuner first starts or whenever the system configuration of processor and/or interactive resources have changed.
The memory tuning task may involve a method for managing the memory of a computer system including storage devices, computer processor power, and a plurality of partitions, where the memory, storage devices, and computer processor power of the computer system are distributively assigned among the plurality of partitions, and where the computer system uses the storage devices assigned to each partition to supplement the memory assigned to the respective partition. This method involves, for each of two or more partitions of the plurality of partitions, providing a first measurement related to the frequency with which storage devices assigned to the respective partition are used to supplement memory in the respective partition. Then, a second measurement is provided that is related to the efficiency with which the storage devices assigned to the respective partition supplement memory in the respective partition. Then, a third measurement is provided that is related to the rate at which the computer processor power assigned to the respective partition is used. Next, a memory pain value is calculated representing the degree to which computing performance in the respective partition is being degraded due to a lack of memory, the memory pain value being directly related to the first measurement and indirectly related to the second and third measurements. Finally, the first value is stored.
The memory tuning task may also involve a second method for managing the memory of a computer system having a plurality of partitions, where the memory of the computer system is distributively assigned among the plurality of partitions. This method involves obtaining, for each of two or more partitions of the plurality of partitions, a value that describes the degree to which computing performance in the respective partition is being degraded due to a lack of memory in relation to the other partitions of the two or more partitions. Then, memory is allocated among one or more of the two or more partitions so that future values expected to be obtained for each of the two or more partitions after the implementation will be more evenly distributed among each of the two or more partitions than the current values are distributed among each of the two or more partitions.
In the memory tuning task, the tuner may first determine the relative memory pain rates for each partition to be tuned. Then the tuner allocates memory among these partitions so that future relative memory pain rates expected to be obtained for each of the
Mendoza Dwight Anthony
Schnee David Z.
Barsa Consulting Group, LLC
Brown & Raysman Millstein Felder & Steiner LLP
Ellis Kevin L.
LandOfFree
Method and system for automatically measuring resource needs... 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 system for automatically measuring resource needs..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for automatically measuring resource needs... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3335766