JPH04502527A - Computer system memory performance improvement device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] FIELD OF THE INVENTION Computer System Memory Performance Improvement Apparatus The present invention monitors the data retrieval efficiency of computer system memory and data set in response to memory operation conflicts detected in the data system memory. Concerning computer system memory that changes storage location, especially computer The present invention relates to a system memory performance improvement device.

assignment ] Efficiently managing the storage of datasets on computer system memory is , is a problem in data processing systems. large number of computer systems has a hierarchy of data storage devices used within it. These data records storage device from an offline magnetic tape cartridge data storage system. Direct access storage that is connected directly to your computer system and available online and of datasets that are ultimately frequently read by computer systems. This extends to cache memory, which is high-speed online memory intended for use. child The difficulty with the hierarchical structure of data storage is that it is difficult to Dynamically allocate to these various data storage devices and retrieve data from these data sets. Ensuring that frequent data retrieval is compatible with its storage location and data storage It is to measure the wise search efficiency.

Place the most frequently retrieved datasets in cache memory and the least frequently retrieved datasets in cache memory. Data sets are searched without offline magnetic tape cartridge data storage location on the device is advantageous for performance purposes of the computer system.

According to research conducted in the field of data processing, in 1971, data processing systems The intermediate facility has 1.2 gigabytes of direct access data storage capacity. These data storage devices had 450 data sets.

A 1984 study found that a typical or intermediate installation had 92 gigabytes of direct access. The device has an access data storage capacity of 44,000 data sets internally. I had paid it. Direct access data storage in a typical computer system There has been continuous and notable growth in capacity. However, direct access Data storage utilization has increased from 80% in the late 1960s to today. This has fallen to about 55%. Then data storage in computer systems There is a reduction in efficiency when using direct access data storage within a hierarchy of devices. Therefore, the cost of memory for direct access data storage devices increases significantly. computer 10 in one computer system equipment to manage the data system memory. Requiring approximately one data management employee for every gigabyte of data managed. In other words, it is a law commonly used in the field of computer systems. This de A data administrator uses direct access data storage within a computer system. Balancing input/output loads and arranging datasets to improve efficiency in performs cache tuning duties. Therefore, data storage Any improvement in memory management is necessary to manage computer system memory. Reduce the number of data management employees.

solution This computer system memory performance improvement device solves these problems. , technological advances are achieved in this field. This device is compatible with computer systems. Monitor system data storage performance, check for memory operation conflicts, and by directing the relocation of datasets to other segments of computer system memory. utilizes knowledge-based (expert) systems to resolve these conflicts. do.

A subordinate device is an external device used to manage the data storage of a computer system. Many databases used by Spurt System software Contains. One element provided by this device is a variety of data storage devices and of -sets, giving a description of their interconnection in a computer system. Data storage device configuration data. The second element represents data storage management functions. A knowledge database containing a set of functional rules. These rules vary The operating characteristics of data storage devices and the requirements for computer system memory It represents the measures that should be taken to provide improved functionality.

An expert system runs on these two databases; That is, a level of performance comparable to that of human experts in the field of application. A computer program that uses clearly written knowledge and computational reasoning techniques to achieve It is a computer program. Expert systems are the basis of expert systems. Contains an inference engine that executes rules with underlying intelligence. The inference engine is , a virtually infinite number of rules or conditional statements chained together in various ways allow yourself to be The inference engine analyzes the rules and the flow of data through those rules. Manage. The expert system then uses the computer system's data. computer system to analyze the performance of data storage devices on a dynamic basis. The configuration database as well as the data that comes from monitoring the actual performance of data storage devices using data stored in databases and knowledge databases.

An expert system is a component that attempts to access a common data storage device. computer system due to multiple computers in the computer system Identify memory operation conflicts, such as data storage performance degradation. expert The system is identical to the data set stored on the data storage device involved in the conflict. Identify behavioral conflicts. Once a dataset is involved in an operation conflict, Once confirmed, the expert system will consider alternative methods for these datasets. Determine memory storage locations and move these conflicting datasets to alternative data storage locations. Invoke various software routines to transfer data to the computer. These competing designs Relocation of data sets resolves memory operation conflicts and Improve searchability of stored data. Conflict detection and dynamic real-time The data storage system of a computer system is devices operate in a more efficient manner and the data stored on these data storage devices Searchability is greatly improved without the need for data management personnel. Therefore, This computer system memory performance improvement device is monitor and correct the performance of data storage devices that have been

Brief description of the drawing "Figure" shows the configuration of a computer system memory performance improvement device in block diagram form. show.

FIG. 2 shows, in block diagram form, an apparatus for improving computer system memory performance. 2 shows the configuration of cache tuning and DASD performance optimization device elements.

Figures 3 and 4 show cache tuning of a computer system memory performance improvement device. 3 shows a group of elements including a DASD performance optimizer element and a DASD performance optimization device element.

detailed description This computer system memory performance improvement device improves computer system memory. Monitor your computer system memory performance and use this command to maximize performance. Performs the function of managing computer system memory. This feature is a knowledge base (e.g. This is accomplished using the Kispert) system. The knowledge base system is computer system memory to improve computer system memory performance monitors the activity of various data storage devices in the Review and take steps to resolve these conflicts. This device is a data Used by expert system software to manage storage devices Contains many databases.

Data processing system architecture Figure 1 shows a typical data processing system in block diagram form. The system includes a computer system memory performance improvement device. This computer The data system consists of mainframe 101, which is 18M3084, Amdahl 5860 or NAS 9000 series computer Any of many large computer systems, such as a computer. these Each mainframe computer operates on the order of 10,000 instructions per second. and typically include large amounts of disk storage memory. The data record Storage capacity is typically on the order of 30 to 100 gigabytes. For explanation, The mainframe 101 is an IBM 3 that runs an operating system 102. Assume that it is a 084 type computer. The operating system 102 is Typically an MVS/XA operating system. These products are industry is well known and does not require further explanation here. Mainframe 101 is connected to the mainframe 10 via a plurality of channel processors 110-119. 1 and an associated data storage device. For illustration, a large number of disk storage Devices 150-0 through 150-15 are shown in FIG. These disk drives Each of the devices may be an 8380E disk drive or a number of similar drives. This is a commercially available device. A typical configuration has 16 disk drives. data channels 140-0 to 140 to which the locations 150-0 to 150-15 are associated; -15, the device controller 1, typically an 8890 type controller. 30. The device control device 130 communicates with the device control device 130. A plurality of disk drives are connected on the channel bus 120 connecting the channel processor 110. Performs the function of multiplexing data going back and forth from device 150-0 to device 150-15. .

The computer system is shown as devices 110 through 119 in FIG. A plurality of channel processors may be provided. Channel processor 11 0 is connected to the mainframe 101, and the disk storage is accessed from the mainframe 101. Allows access to devices 150-0 to 150-15. This block diagram is Although the configuration of a general computer system is shown, the applicability of the applicant's invention is It should not be construed as limiting. In particular, various memory configurations 101 and various data storage devices. Shown in Figure 1 disk drives are exemplary, and many computer systems , rely on tape drives to provide data storage. Shown in Figure 1 A computer system that uses cache memory, which is part of the device control unit, 160 are included. The memory performance improvement device 103 is connected via a data link 160. Similarly to the operator terminal 170 connected to the main frame 101 via Included in frame 101.

Memory configuration and management From a data management standpoint, the computer memory shown in Figure 1 is a disk storage A subsystem 180 consisting of 16 volumes 150-0 to 150-15; be considered. Each volume in this subsystem can contain multiple data sets or Contains data files. These datasets or data files are on the corresponding volume in an orderly manner by dividing the volume into subelements. Stored. For example, on an 8380E disk drive, each volume has 177 It consists of 0 cylinders, each cylinder containing 15 data tracks. Each data A truck can store 47 kilobytes of data. In operation, the mainframe The system 101 is connected to a data volume example 150-0 containing the requested data. service by sending a data retrieval request via the channel processor 110. access data stored on data volumes within the system 180;

This data retrieval request is sent by channel processor 110 to channel bus 12. 0 to the device controller 130, which in turn sends the instruction Manage data retrieval from the indicated data storage devices 150-0. This data inspection The search information is stored by the device controller 130 on tracks of the disk storage system. The rotation of the disk drive 150-0 is monitored to determine the starting point of the data set. This is achieved by looking at it.

- Whenever a disk in a disk storage system is When rotated all the way to the bottom of the read/write head of the read/write head of the disk storage system, the data From the disk drive 150-0 to the device controller through the data link 140-0. The information is sent to the station 130. The device control device 130 is connected to the main frame 101. search data to be sent via channel bus 120 and channel processor 110. Store temporarily.

What is clear from this arrangement is that the mainframe 101 handles various data retrieval requests. to the device controller, where these requests are sent to the device controller when the data set is available. The above transfer is an asynchronous data transfer. Ru. If too many data retrieval requests are made via one route i.e. if a specific volume or small subset of volumes When demands are concentrated, the performance of computer system memory is significantly reduced. various Data detection is performed through various channel processors, channel buses, and disk drives. It is beneficial to equalize search demands.

Data retrieval across various volumes and across all subsystems Distributing requests greatly improves memory performance. data set in one Move a volume to another volume, or combine a volume into a subsystem. system to other subsystems to improve memory performance. There are many choices available to help you. Also, for this patent disclosure, [String J and “ Keep in mind that the terms "subsystem" are interchangeable. It should be done. Another such improvement is the use of keys such as cache 160. The use of cache memory, which allows frequently accessed volumes to be Enables data to be loaded into the cache while data retrieval is performed. do. Caching based on the dataset stored on the volume The volume is selected such that it is appropriate. Use cache memory I/O access to data is improved by speeding up data transfer time. profit from cash. To ensure data integrity, all licenses The write operation causes data to be written directly to disk and therefore this These operations do not benefit from cache memory. suitable for caching Volumes with high read speed, high activity (liveness) and I/O access concentrated on a few tracks rather than distributed throughout It is something that has. A volume that satisfies these conditions is likely to have a high hit rate. Therefore, an I/O read will often find the desired track in the cache and Spending the time required to retrieve the dataset from disk as mentioned in There's no need. Often, subsystems do not even contain any cacheable volumes. activity on a cacheable volume is Responsible for only a portion of the total input/output activity of the system.

Improved cache performance Head to a specific cache volume for full cache effectiveness Reorganizing subsystems in a way that increases the amount of input/output access is possible. The key issue is data sets that degrade cache performance. at the dataset level to separate cacheable datasets from Cache memory is used to reorganize subsystems and The goal is to create a full volume suitable for storage within. Data set organization Remodeling involves moving a huge number of datasets to achieve It is clear that this can cause a complete remodeling of the stem volume. It is. Instead of a complete reorganization of the subsystem, a better solution is to There is a conflict between cached data and data that is unfavorable to the cache. It is about identifying the location on the subsystem. Conflicts are good on the same volume. There exists bad cache data and bad cache data. Sunawa If you want to benefit from caching data with good volume, Bad data actually reduces overall performance if cached for would cause more interference in the cache. computer system The system memory performance improver locates the conflict and then uses the system to resolve the conflict. Move the least number of datasets. With this solution, a small number of moves can improve performance. Provides significant improvements and better utilization of cache memory.

Which datasets are possible candidates to cache, and which data subsystem input to determine if a set should not be stored in the cache. Output activity is monitored over a period of time, and the data is recorded on the source. This recorded data consists of input/output activity (data (number of input and output operations for the data set), read rate (percentage of read operations) ), and the degree of locality of reference. The degree of locality of reference is determined by the time period to be monitored. the dataset divided by the number of tracks handled by input/output over Defined as the number of inputs and outputs for

Along with these numbers, the dataset name, size in bytes, and the move is appropriate. Contains values called reason codes that flag invalid datasets. child The data includes the datasets to be moved and the target locations for these datasets. With this computer system memory performance improvement device, used.

Memory performance improvement device The present computer system memory performance improvement device comprises - a set of integrated data storage management systems; This function is used to limit the performance of direct access data storage subsystems. knowledge or expert systems for monitoring and automatic assistance in management. use the technology of the system. Figure 2 shows the computer system memory performance improvement device 1. This shows the basic components of 03.

The memory performance improvement device 103 uses the MVS operating system of the mainframe 101. It interfaces with and runs on system 102. MVS Operator The management system 102 includes device management, file management, test management, processor management, communication management, and computer system memory performance improvement device 103 and many other management functions.

To clarify the terminology, various aspects of the knowledge system are considered. knowledge system Systems interpret the encoded knowledge of human experts stored in knowledge bases. A code that emulates human reasoning by using an inference engine to It is a computer-based system. If the domain of the knowledge base is sufficiently limited and that a sufficiently large volume of knowledge is accurately symbolized in the knowledge base. If the knowledge base has performance that matches or exceeds that of human experts, can be achieved. In such cases, the knowledge system It becomes the stem.

The first step in building an expert system is to input the accumulated knowledge of human experts. and encode it into a machine-readable expert system language. That is. Details of the specific implementation of this symbolization step are available from selected experts. This is largely a function of the unique syntax of the system's programming language. -For example, One of the programming languages for expert systems such as PROLOG is It is the text 'Programn+ing In Prolog'' (11, P, C1ocksin and C, S, Mell ish, Springer Verlag Inc., New York, New York 1981) It is described in Essentially, an expert system uses rules and methods to solve a particular problem. instructions as to how the rules should be applied to the facts available to determine the Contains a set of

In this computer system memory performance improvement device 103, an expert system technology is used to monitor computer system memory performance . The facts gathered through this monitoring operation are then compiled into computer system notes. computer systems, as well as the data stored on them, to improve their performance. used to identify modifications to the organization of system memory. Figure 2 shows the computer Various routines and subsystems realized by the data system memory performance improvement device 103 Some of the stems are shown.

It includes a cache volume generation function 105 and a DASD performance optimization function 1. 06 is included. The cache subsystem is often the last to If there are no volumes on the DASD subsystem suitable for The number of Ilo's for the volume being It will be in a state where only the ratio of the number of barrels will be revealed. Cache volume generation 105 creates a volume (205) on a subsystem suitable for caching. Ru. Cache volume generation 105 is based on the nature of cache memories 160-169. 201), and whether cache candidate datasets are cacheable or not. (202). Cash Volume generation 105 analyzes conflicts in datasets and separates good candidates from bad ones. Recommends the movement (203) of the dataset to separate it from the cache, thereby Concentrate candidate datasets onto a small number of cacheable volumes.

The cache volume raw Fli, 105 is called by the scheduler 109. Analyze one DASD subsystem each time and migrate the minimum number of data sets. (204) to try to achieve maximum effect. DASD performance optimization function 10 6 to improve the performance and utilization of DASD devices 150-0 to 150-15. create a dataset move recommendation. The DASD performance optimization function 106 performs preprocessing. Datasets and subsystems with over-used volumes , move to underused volumes and subsystems. Optimal DASD performance function 106 controls the activity in DASD devices 150-0 to 150-15. (211)L, monitor over- and under-used volumes and subsystems. Identify potential ASD performance problems caused by the system (212>).

The DASD performance optimization function 106 then adjusts the volume to avoid performance problems. The data sets are relocated (213) between the data sets.

Scheduler 109 controls the operation of all functions and allows activities to occur at any time. Decide whether The parameters used to establish the schedule are based on facility regulations. performance, peak processing times, list of functions to be performed, and user-provided constraints. Day evening Collection 171 is performed on a pre-scheduled basis (e.g. hourly totals). Record I10 activity for all devices based on the In addition, detailed Volume attility data and track data can be recorded. data collection The collection 171 produces statistical information required by the remaining modules of the system. Ru. Database 108 includes system configuration information as well as statistical activity data. Contains information.

Those skilled in this technology use the term "database" to refer to "knowledge". When it is considered that both "configuration" and "configuration" are included in the general container of data, I would appreciate what you say. The database 108 is stored in the disk drive 15 0-15, but the mainframe computer 1 01 computer memory performance improvement device 103. Dew.

Cache volume generation Cache volume generation (CVC) module 105 This is one of the performance tuning routines of the system memory performance improvement device 103. Cat The cache volume generation module 105 analyzes the cache subsystem and Recommendations to improve input/output I10 performance in computer system memory Use expert system technology to create.

Due to the difficulties involved in using the cache properly or when the job should be done Due to a lack of data management employees, there is no need to create a box on an appropriate subsystem for caching. If the number of ILOs for a volume that is empty or cached is For a while, it is assumed that only the proportion of the total number of Ilo to the system is assumed. It will be the last time. The purpose of the cache volume generation module is to cache This can be done by attempting to create a volume on the appropriate subsystem. It is to correct the condition.

Most preferably, datasets on subsystems benefit from caching. However, the cache operates on a volume-based and dataset is on a volume that would be detrimental to I10 performance when cached. Therefore, these datasets are not cacheable.

The job of the cache volume generation module is to Identify where you are competing with unsuitable datasets to replace good candidates with bad ones. is to recommend moves that separate cache candidates from their complements, and therefore Concentrate the data on a small number of cacheable volumes.

Cache volume generation module achieves maximum performance improvement with minimum number of moves Strive to do so.

Cache volume generation overview The way the cache volume generation module 302 operates depends on which subsystem The first step is to determine whether the datasets on the system need to be cached.

Based on the classification of good cache candidates, the cache volume generation module The module 302 then determines which volumes are good cache volumes. , which ones are bad, and which ones are good datasets that are competing with the bad datasets. Determine if it is a cut. Volumes are good, bad, and contention, respectively. It is classified as The number of cache volumes requested is then determined by the data storage Depends on how many pieces of movement are required to generate this number of cache volumes. will be determined. Then the contention volume is processed and the good data set is separated from the data set. Finally, cache volume generation module 3 02 can create enough improvement in the cache candidate volume with 1 or 2 moves Run the last bus to check if the

Cache volume generation module The cache volume generation module 302 is called by the scheduler 109. Contains a functional entry point that can be accessed. The cache volume generation module 302 Submodules 303-308 as shown in FIG. 3 are then executed. Kya When the disk volume generation module 302 comes to the exit, it knowledge database 108 for performing a list of dataset transfers and according to recommendations; The operating system 102 is called to update the .

On entry, a pointer to the subsystem record for which the subsystem is parsed. can be conveyed. This record contains subsystem information and is also a volume record. Contains a pointer to a list of volumes for each volume in the subsystem. It is. Each volume record contains volume information and a list of dataset records. Contains a pointer to the data set for each data set on the volume. be. Each dataset record contains information about dataset activity. Contains.

Initialization The initialization module 303 specifies any data formats that need to be generated. Performs any initialization required by the cache volume generation feature, including conversion. Ru. Initialization module 303 also performs data storage with scheduled migration. Search for the item. These datasets are transferred when the data is collected for analysis. The move has not been completed, but any moves recommended by the current analysis have been performed. By the time they arrive, these movements have been accomplished. This means that the state of the machine may vary from the time the data is collected to the time the dataset move recommendation is executed. and the dataset move recommendation is executed after the scheduled move has already been performed. This means that it is based on a system that allows you to discern whether the law is legal or not.

Initialization module 303 inputs data to reflect this state of the system. change.

Determine cache candidates The purpose of cache candidate determination module 304 is to determine which data set on the subsystem. The next step is to determine whether the set should be placed on the cache volume. Firstly , the routine “Dataset Classification” 311 is called when the dataset is Depending on how you commonly perform data sets when is classified as inert. The routine ``Determine Cutoff'' 312 is then called. Ru. This routine uses cache candidate data on the subsystem against the cache capacity. Determine if there is too much data, and if so, determine the amount of cache candidate data. Determine a cutoff that reduces cache capacity to the best amount of data set it can handle. Set. The Dataset Reclassification routine 313 then determines the cache candidate data. reclassify the data set until it falls below the cutoff and create another cache sub. Flag them for possible movement into the system.

Classify the dataset The dataset classification routine 311 classifies the datasets on the subsystem as good ( good) cache candidate, bad cache candidate, or inactive cache candidate. (inactive) data set. Classification is an act of a dataset. Some simple rough estimates based on activity, lead percentage, and LORM. It is done in a unique manner. (LORM is the locality of reference measure (Locali ty of Reference Measure), which is per second divided by the number of unique tracks accessed over the period viewed. equal to the number of Ilo. ) data set as a good or bad cache candidate. Similar to classifying datasets using keywords, the dataset classification module also Reflects the appropriate frequency of caching or caching for a given cache capacity. By selecting the datasets to be scanned, the dataset classification module attempts to include the best cache candidates.

rules The rules used to classify the dataset are shown below, where x, y and Z are optional due to specific device characteristics and user needs This is a variable that was selected based on

What if 1. If there were no activities on the dataset At that time, the dataset is inactive.If 1. data Set glue %〉X cutlet 2, LORM>y cutlet 3. If I10 speed > 2 seconds If the dataset is good then 1. (a) Lead% x or (b) LORM<y or (c) I10 speed x z per second and 2. If there are some activities on the dataset At that time, the data set has a bad cutoff decision The cutoff determination routine 312 is determined by the data set classification routine 311. Which cache candidates are identified are cached based on cache capacity The responsibility is to decide whether it should be done. from the best cache candidate Data attributes like handwriting, activity, lead percentage, and LORM Based on the Step down the list of cache candidates until you select one for the available cache capacity. We will process it. This means that the cutoff determination routine 312 is cached. Deciding that there should be a cutoff for the dataset above the cutoff It is a place to do.

In determining where a cutoff exists, the cutoff determination routine 312 Also, look at the current cache data and the read hit rate of the cache and devices. ing. With this method, if the cache is overloaded, some data set is moved from the cache volume and away from the subsystem. flagged as a possible move. Similarly, if cache Even better if it works better than expected and can handle more data than expected. Some cache candidates are moved onto the cache volume. If cut Off decision routine 312 determines whether the given cache capacity is available on the subsystem. If you decide that you can handle more cacheable data than you can handle, then The off-off determination routine 312 determines whether a subsystem has cache that is below utilization. Add a flag to the system.

Dataset reclassification Using the cutoff determined by cutoff determination routine 312, the data The dataset reclassification routine 313 uses a cutoff as a bad dataset. Reclassify datasets that fall below the threshold and move them to other cache subsystems Add flags to them as possible.

volume classification Volume classification routine 305 determines whether a volume is a good cache candidate or a bad cache candidate. Is it a bad cache candidate or is there a bad cache candidate on the volume? Based on whether there is good cache data that is competing with The volume of a system can be classified as good, bad, or competitive. nflict) volume. The volume classification routine 305 Use the classification of the dataset to determine the classification of the volume. Moshiboli A volume is classified as good if it mainly consists of good datasets. be done. If a volume consists entirely of bad datasets, then the volume volume is classified as bad volume.

If there is good data on the volume, but there is also a large amount of bad data. If so, there is contention between good and bad data and the volume is above the contention. Classified as a volume.

where n is the characteristic of the device being monitored and the user needs. is a predetermined threshold selected as a function of

What if 1. If there is no good dataset on the volume At that time the volume is bad What if 1. If there is good data on the volume and 2, the amount of bad data is has a value greater than n of the total activity of the volume, then If the volume is a competing volume then 1. good on volume bad data exists, and 2, the amount of bad data is the total volume activity. has a value smaller than n of , then At that time the volume is good Cache volume generation The cache volume generation routine 306 determines how much cache volume determine whether a volume should exist on a subsystem and if it is good enough Responsible for creating volumes if they do not already exist. At first, The cache volume number determination routine 314 uses a predetermined device (device). cache requested based on activity level and space constraints Used to determine volume number. of the requested cache volume. The number determines how much better cache volume should be generated. is compared to the number of good cache volumes present to determine Ru. The resolution strategy determination routine 315 determines how to resolve conflicts on conflicting volumes. Decide what to resolve. The solution strategy determination routine 315 determines which data has a bad solution strategy. By looking for the volume you are trying to move the tasset from , in the end we find a volume that will be a good volume. This is which In this way, the resolution strategy determination routine 315 is The question is whether to generate a number of programs. Move bad datasets with a what-if solution strategy If there are not enough contention datasets, cache volume generation routine 3 06 is a volume that requires a minimum number of moves to change to a better volume Find out. The dataset placement routine 316 performs a cacheable volume where to relocate the dataset that needs to be moved to generate the used to make decisions.

Determine the number of cache volumes The cache volume number determination routine 314 determines how many cache volumes there are. determine whether the volume should exist on the subsystem. This decision was made after Space requests for cached candidates, access for cached and non-cached candidates expected read hit rate for caches and devices, and subsystems. Based on the total amount of space. The purpose is to address device contention issues and non-caching Similar to trying to minimize space issues on device volumes, queuing is not a problem and space requests for cache data are not met. So, it is important to have enough cache volume.

Conflict resolution Conflict resolution routine 307 resolves any remaining conflicts by resolving conflicts on the volume. Handle conflicting volumes. Contention is caused by caches that exist on the same volume. Occurs where there is cached data and large amounts of non-cached data. the volume The system uses caches because bad data degrades cache performance too much. It is impossible to do so. Conflict resolution routine 307 determines on the resolution trajectory , then execute the strategy.

The resolution strategy is determined by the resolution strategy determination routine 317 and is one of the following: It is possible.

That is, one of the following actions can be selected to resolve the conflict.

1. Move good datasets to good volumes. 2. Move bad datasets to bad volumes. move to a new volume, or 3. A good volume of movable data sets with enough activity Rescue as many good datasets as possible by moving to

- Once a solution strategy has been determined, the dataset placement routine 317 used to decide whether to move identified datasets to this location.

Solution strategy determination routine Given the contention volume, the best way to remove contention must be determined. do not have. That is, it is the best way to separate good data from bad data from each other.

It is possible for good data sets to be relocated away from the volume, and it is possible for bad data sets to be relocated away from the volume. It is also possible that the new data is relocated away from the volume, or It is also possible that as many good datasets as possible are rescued.

Salvaging a good dataset is used as a last resort. Where, A bad dataset, perhaps due to an immobile dataset or the number of movements caused. It is impossible to move the data to separate it from the volume, and it is impossible to move the data to separate it from the volume. It is also not possible to move all of the tassets to detach them from the volume. child Under these circumstances, a good data set with large activity and mobility is By moving the data sets, good data sets can be rescued.

What is actually meant about bad datasets on contention volumes is that bad A selected subset of the dataset. good volume, activity volume activity is less than n due to bad data set. It is still possible to have bad datasets on the system.

Then, in order to convert contention volume into good volume, The tasset must be moved to reduce bad activity to this level. No. The selected bad datasets were then moved from the volume Reduce the value of bad activity to less than 20% of volume activity if This results in a data set that allows

Factors that influence the decision which resolution strategy to use are described below.

Each of these factors is represented as an attribute of the contention volume, and its value depends on the input data. derived from data and other attributes.

The derivation of the values of the above attributes and the attributes from which these values are obtained will be explained in detail in a later section. be done.

rules Rules to consider using these factors to resolve conflicts are provided below. Ru.

What if 1. There are no immovable bad datasets, and 2. If there are some completely immovable bad datasets, then At that time, the solution strategy is "error" What if 1. There is no stable good data set, and 2. If there are some perfectly immobile good datasets, then At that time, the solution strategy is "error" What if 1. Good data and bad data are equally immovable, and 2. Some stable and good data sets exist, and 3. If there is no immovable bad data set, then At that time, the solution strategy is "error" What if 1. Good data and bad data are equally immovable, and 2. There are some bad datasets, and 3. If a good immovable data set does not exist, then At that time, the solution strategy is "error" What if 1. Good data and bad data are equally immovable, and 2. Some perfectly immovable good data sets exist, and 3. If there is no completely immovable bad data set, then At that time, the solution strategy is "error" If △1. Good data and bad data are equally immobile2, and some are completely immobile. There is a dynamic bad dataset, and 3. If there is no perfectly immovable good data set, then At that time, the solution strategy is "error" What if 1. Good data is more immovable, and 2. Good data is immovable. If the cut does not exist, then At that time, the solution strategy is "error" What if 1. Bad data is more immovable, and 2. Bad data is immovable. If the cut does not exist, then At that time, the solution strategy is "error" What if 1. Good data is more immovable, and 2. some completely immovable. If a dynamic bad dataset exists, then At that time, the solution strategy is "error" What if 1. Bad data is more immovable, and 2. some completely immovable. If a good dynamic dataset exists, then At that time, the solution strategy is "error" What if 1. We proceed to move all of the good datasets, and 2. We do not move bad data willingly, and 3. There is no fixed good data set, then At that time, the solution strategy is "move - good" What if 1. We are willing to move forward with bad data. , and 2, we don't move all of the good datasets, or One 3. There are no immovable bad data sets, or 4. The time required to move the bad data is Whatever is required is good enough data to justify a special move. If it exists, then the solution strategy is "move - bad" What if 1. we Proceed to move all of your good datasets, and 2. We do not move bad data willingly, and 3. There are some stable good datasets, and 4. Good data is fully active, or 5. Good data set is completely immobile. does not exist, then Then the solution strategy is “move-good” What if 16 we move forward with bad data , and 2, we don't move all of the good datasets, or One 3. There are some persistent bad datasets, and 4. Is the good data active enough? 5. Is it necessary to move the bad data? There is enough undesirable good data to justify any special moves required. exist and 6. There is no completely immovable bad data set, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. There are no permanent good data sets. 4. There are no fixed bad data sets. 5. Good data is If you require more than one movement, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. There are no permanent good data sets. 4. There are no fixed bad data sets. 5. Bad data is If you require more than one movement, then At that time, the solution strategy is “Move - Good” What if 1. We move forward with good data move all of the set, and 2. We are willing to move bad data, and 3. There are no permanent good data sets. 4. There is no permanent bad data set. 5. Good data and bad data. data requires the same number of moves, and 6. Good data has more than one activity, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. There are no permanent good data sets. 4. There is no permanent bad data set. 5. Good data and bad data. data requires the same number of moves, and 6. Bad data has higher activity, then At that time, the solution strategy is "move - good" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There is no fixed bad dataset, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. There are no permanent good data sets. 4. Does not exist? If some persistent bad dataset exists, then At that time, the solution strategy is "move - good" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Is good data active enough? 6. Good data is more active than bad data. immovable and 7. There is no completely immovable bad data set, then At that time, the solution strategy is ``Move - Bad J What if 1. We move forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Is good data active enough? 6. Bad data is more active than good data. immovable and 7. There is no perfectly immovable good data set, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Good data is active enough, or 6. Good data and bad data are equal. be very steadfast, and 7. There is no perfectly immovable good data set, and 8. There are no completely immovable bad datasets, and 9. Good data requires more movement, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Good data is active enough, or 6. Good data is bad data, etc. be very steadfast, and 7. There is no perfectly immovable good data set, and 8. There are no completely immovable bad datasets, and 9. Bad data requires more movement, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Good data is active enough, or 6. Good data and bad data are equal. be very steadfast, and 7. There is no perfectly immovable good data set, and 8. There are no completely immovable bad datasets, and 9. Good data and bad data require the same number of moves, and 10. Good data has higher activity, then At that time, the solution strategy is "Move - Bad" What if 1. We are moving forward with good data. move all of the set, and 2. We are willing to move bad data, and 3. Some persistent good data. set exists, and 4. There are some persistent bad datasets, and 5. Good data is active enough, or 6. Good data and bad data are equal. be very steadfast, and 7. There is no perfectly immovable good data set, and 8. There are no completely immovable bad datasets, and 9. Good data and bad data require the same number of moves, and 10. Bad data has higher activity, then At that time, the solution strategy is "move - good" What if 1. Good day we move on Do not move all of the tassets, and 2. We don't move bad data willingly, then At that time, the solution strategy is "rescue - good" What if 1. some completely immovable A good dataset exists, and 2. If there are some completely immovable bad datasets, then At that time, the solution strategy is "rescue - good" What if 1. some steadfast good de data set exists and 2. There are some persistent bad datasets, and 3. Good data is not active enough, then At that time, the solution strategy is "rescue - good" What if 1. We are moving forward with good data. move all of the set, and 2. We do not move bad data willingly, and 3. Some stable and good data sets exist, and 4. (some perfectly immobile good dataset exists, or 5. If good data is not active enough) At that time, the △ solution strategy is “rescue-good” What if 1. We are willing to move forward with bad data. , and 2, we don't move all of the good datasets, or One 3. There is no immovable bad data set, and 4. It is difficult to move bad data. Sufficiently good data exist to justify any special moves required. If not, then the solution strategy is "rescue - good" What if 1. we 2. We are willing to move the bad data, and 2. We are willing to move all of the good data sets. do not move, and 3. There are some persistent bad datasets, and 4. (Good data is not active enough or 5. To justify any special moves required to move the bad data. There is no sufficiently undesirable good data, or 6, some completely immovable evil if there is a good dataset) then At that time, the solution strategy is "rescue - good", immovable and bad This attribute is used if there is any immovable bad data (in the already selected subset). If so, it is true.

It's good to be immobile This attribute is useful if some good dataset on the volume is immovable. , is true.

move evil forward This attribute is useful if we are willing to move bad datasets from the volume. Raba is true. i.e. the number of moves required to move the bad data set. if the number is less than or equal to the number of moves we are willing to perform. , is true.

The value of this attribute is the value of two attributes depends on the value of

move on and move on This attribute is used when a decision is made to move all of the good datasets from the volume. If so, it is true. For bad datasets, move the bad dataset The number of moves it takes to move the dataset to determine whether is compared to the number of moves selected to perform. As for a good dataset However, the dataset is based on the number of moves rather than on the number of moves selected. It may be a movement candidate selected based on the Because of low activity Because there are some good datasets that are not worth moving. mutual affection This attribute is useful if all of the good datasets are worth moving and required. True if the number of moves selected is less than or equal to the number of moves selected. Ru.

This attribute is based on these attributes, i.e. depends on the value of

Items that require more movement than The value of this attribute makes it easier to move good datasets through volume. or whether you need more movement to move a bad data set. Indicates whether movement is required.

The values it can take are It is. This attribute is depends on the value of

something with more activity than The value of this attribute is determined by whether the activity on good data is greater than on bad data, etc. or the bad data has more activity than the good data. It shows whether there are any. Its possible values are bad Good data active enough This attribute indicates that if the combined activity of good data is immobile data True if the set is active enough to consider moving. child The decision is about the total amount of good activity on the subsystem. Get good data activity on the system and also more cached data Based on importance.

The value of this attribute is activity classification, and More or less depends on the importance of getting cached data.

rules The rules used to determine whether good data is active enough are listed below. shown.

What if 1. Good data has moderate activity and 2. It is very important to obtain more cache data, then If then good data is active enough 1. Good data is expensive activity, and 2. It is very important to obtain more cache data, then At that time, if good data has enough activity 1. Good data is If you always have high activity, then At that time, good data is more stable than having sufficient activity. The value of this attribute is more constant for either good or bad data than the other. It shows whether there is. At that time, the most stable dataset of data in one form is If it is more immobile than the most immobile dataset in another format, then Data is more immovable than anything else. The possible values for this attribute are This attribute is at least completely immovable and should not be moved under any circumstances. is true if one data set exists. For example, what is the VTOC? It must never be moved for any reason.

some good and completely immovable thing This attribute is true if some good datasets are completely immobile. .

It's good to be reluctant enough When there are good data sets that are not candidates for movement (their activity is This parameter can be used to move bad data sets from the volume These reluctances are sufficient to justify any extra travel required to Indicates whether a suitable dataset exists. The one we choose, we proceed Look at the good dataset you are trying to move, move it, and end up on the bad volume. The last thing you want to do is leave behind a good data set. This attribute If the combined activity of the good data set is involuntary, then the bad Any extra movement required to move the dataset from the contention volume is true if it is sufficient to justify it.

number of moves to move evil The value of this attribute determines the amount of migration required to move a bad data set from the contention volume. Indicates the number of movements. The number of moves is determined by the dataset placement module.

Number of moves to move good This attribute is required to move all of the good data sets from the contention volume. gives the number of moves to be made. Number of moves determined by dataset placement module be done.

willing move When creating movement recommendations, the procedure is to minimize the number of movements. each conflict For volumes, basically about the overall amount of good data on a subsystem. There are a number of moves that can be made based on good data activity. That's it However, where there is a lot of good data, more movement is desirable; Where only good data exists, less movement is desirable. good data Similar to activities, the importance value of getting cached data over attributes is , will be closely examined. If more cache data is added to the cache volume If it is important to place the It becomes.

rules The rules used to determine the number of moves are shown below.

What if 1. Good data set activity indicates good data on subsystems is X% of the total amount of 2. Obtaining more cache data is not important, then Then the number of moves we are willing to make on this dataset is Ru What if 1. Good data set activity indicates good data on subsystems is X% of the total amount of 2. It is important to get more cache data, then Then the number of moves we are willing to create is 1.5*X15 What if 1. Good data set activity indicates good data on subsystems is X% of the total amount of 2. It is very important to get more cache data, then Then the number of allowed moves is 2*X15 more cache data Importance of getting this attribute indicates how important it is to get more cache data. reflects the importance of Of course you get more cache data However, this attribute makes it more important than Indicates whether the item is important or not. It is important to get more cache data The decision as to whether Based on the amount. Possible values for this attribute are: The amount of good data that the cono module expects to end up on a good volume How to measure the amount of good data on contention volume that is easily resolved and add this to the amount of good data already present on the good volume. . This method works harder to get better data onto the cache volume. It gives a very rough measurement that is used to decide whether to work or not.

Used to investigate and predict whether bad data will eventually be cacheable. be done. The easy-to-understand solution that is adopted is that it is not yet cacheable, but However, in the end, the amount of good datasets that can be cached is easily resolved. Assuming approximately the same amount of good data currently exists on the contention volume. Is Rukoto.

rules Criteria used to determine the importance of getting more cached data The rules are shown below, where r and S are user-selected variables.

What if 1. Good amount of data on good volume overall due to activity less than the quality r, and 2, less than r on contention volume that is easily resolved (depending on activity) ) If overall good data exists, then At that time, it is very important to get more cache data What if 1. If (depending on activity) good data is cached between r and S possible, and 2, an overall smaller than r on the contention volume that is easily resolved. good exists At that time, it is important to get more cache data What if 1. If more overall good data than s is cacheable, then At that time, getting more cache data is not important What if 1. There is overall good data less than r on the good volume, and 2. There is a global goodness between r and S on a contention volume that is easily resolved. , then it is important to get more cache data than to get more cache data. Easily resolved conflicting variables to determine the importance of obtaining It is necessary to measure the amount of good data on a volume. The contention volume is also Shimo 1. There are no persistent bad datasets, and 2. We are willing to move forward with good datasets. The number of moves you are going to make with respect to the volume will convert the volume into a good volume. greater than or equal to the number of bad datasets to be moved to If so, it is considered to be easily solved.

Activity classification This module is used to categorize some good data activities be done. Its classification is one of.

Classification is good activity on a subsystem that results from good data being classified. Based on the percentage of the total amount.

rules The rules are shown below, where t, u and V are device characteristics and user is a preselected variable that is a function of Zanes.

What if 1. Good data activity is more than one of the whole fishing wheel activity Yes, then Activity is very high at that time What if 1. Good data activity is greater than or equal to U of the entire fishing wheel, and 2. If the activity is less than or equal to t, then At that time, activity is high What if 1. The entire fishing vehicle has an activity of V or higher, and 2. If the activity is a whole fishing wheel smaller than U, then At that time, the activity is moderate. What if 1. Good data has some activity and 2. If the activity is a whole fishing wheel smaller than V, then At that time, activity is low good data rescue During conflict resolution, the preferred choice to resolve the conflict is the chosen bad dataset. from the volume, or all good datasets from the volume. It is about moving. However, if the data set is immovable or the required With good data activity not high enough to warrant the number of It is possible that neither of these two options can be recommended. In this case, the solution The result of the process is a recommendation that good data on the volume will be rescued. Is Rukoto. Rescue behavior determines whether it is worth moving to a better volume. consists of examining each good data set on itself to determine the to make it cacheable.

If a good dataset is not immobile, then the number of allowed movements is If it's good enough to move, it's worth rescuing. Good data set is immobile, but not completely immobile, then the number of movements allowed is that Not only does it have to be enough to move the cannot be rescued with sufficient activity to warrant movement.

Weekly The above conditions for rescuing a good dataset can be expressed as the rules shown below. can.

What if 1. We move forward with good data sets, and 2. The dataset is not immovable, then move the dataset What if 1. We move forward with good data sets, and 2. The dataset is immovable, and 3. The dataset has enough activity, then At that time, move the dataset Dataset placement The dataset placement routine 316 is used to locate datasets that are known to be moved. used to determine where to place the target. It is the same matching attempts to recommend the placement of datasets on the least active volumes of their type. try. If there is no equivalent volume to which the dataset could fit For example, the dataset placement routine 316 moves the data set to another equivalent volume. an equivalent volume that frees up enough space for the first data set when Try to find another dataset on the volume. as a last resort As such, the dataset placement routine 316 knows when the placement will be easier. If you wish, you can move the data set to another contention volume for later processing. think about it.

Once a placement is found for the dataset to be moved, a recommendation is generated and at the end of the current recommendation list by calling recommendation generation routine 318. It will be destroyed.

Similar to actually placing the dataset, the dataset placement routine 316 , how many times does it take to move a given set of datasets from a volume? Used during the conflict resolution process to determine if more moves are required. This feeling The information then determines which good and bad data is moved from the contention volume. Used when determining whether more movement is required to

Recommendation generation This routine handles the movement of datasets and updates them to the current advisory list. Add to strike.

Final stage optimization Just before the cache generation routine 302 finishes its analysis, the final optimization routine Chin 308 has good volume where one or two moves result in dramatic performance increases. called to try to find a program. The final optimization routine 308 , look at all of the volumes that are good or cache candidates, and set the volume's key One more moveable from volume to increase cache adaptability Or identify a bad dataset with two moderately high activities.

If one person is responsible for more than a certain percentage of the total volume activity, If there is a bad data set on a good volume that can be moved only by When , data set migration from the volume is recommended. (309) Similarly, If more than a predetermined percentage of the total volume activity There is a bad data set on a good volume that is responsible for If possible, then it is recommended to move it from the volume.

DASD performance optimization mechanism DASD performance optimization organization (DPO) module 106 is a computer system This is one of the adjustment analysis routines of the memory performance improvement device 103. Optimal DASD performance The conversion mechanism module 106 analyzes the performance of the DASD device in the equipment, and Expert systems are used to make recommendations for dataset movement that improve performance. using technology of technology and artificial intelligence.

The purpose of the DASD performance optimization mechanism module 106 is to improve the performance and utilization of DASD devices. The objective is to create dataset migration recommendations with the aim of improving That model Joule strives to minimize the number of recommended moves. The recommendation is data collected as a result of monitoring activity and device characteristics. Based on data. The module waits for the problem to occur before doing anything. Rather, it works in such a way as to stop the problem from occurring. Optimal DASD performance The system module 106 identifies equipment that is likely to have a problem or that already has a problem. identify the location (device) and subsystem, and then take the few steps to correct the situation. Create a dataset move recommendation. In this way, the data set is volumes and subsystems to underutilized volumes and subsystems. Ru. The DPO (DASD Performance Optimization Organization) module 402 also provides a cache cache candidate datasets that can be moved to subsystems that are underutilized. Determine where it exists. If necessary, it can also be When the cache of the dataset is overloaded, the data set is Move from.

Any required initial setup is performed by the operating system 102. When the DPO module is executed, the operating system DPO module as a subroutine that conveys all the information you need in a defined format. 402. When the DPO module 402 is finished, it Returning to operating system 102, the operating system performs the necessary Perform any necessary finishing touches.

DASD performance optimization mechanism After some initialization, the DASD performance optimizer module 402 finally under the cache control device by the cache volume generation module 105. Efforts are made to reorganize datasets that have been flagged for move to Melt. Subsystems and volumes with problems or potential problems are identified.

Then, the combination of datasets on these subsystems and volumes errors will be identified and their movement will correct the situation. Identifying overuse Each received data is then processed according to priority, resulting in three selectable data sets. A combination of cuts is selected for each. The processing is done using the best dataset choosing combinations of , deciding where to move them, and including generating appropriate recommendations.

DASD performance optimizer module 402 controls high-level execution flow and to implement a list of dataset move recommendations and to integrate these recommendations into the knowledge database. calls operating system 102 to write to database 108 .

Initialization Initialization routine 404 completes the request process, including reformatting the data if necessary. performs any initialization that is performed. This initialization process includes data storage configuration, each subsystem within the configuration and each subsystem for which a record exists in the database. information from the database to create an internal record for each volume in the Including reading. Initialization routine 404 also includes data storage system Enter pools and hosts that share information about each of the volumes stored within them. create objects for each of the existing compatibility classes. The compatibility class is Combination user boolean, set of common hosts and cacheable format . If a move is planned for the data set, the initialization routine 404 data to reflect the state the system would be in if the move was performed. change the data. The reason for doing this is that the recommended move is already a planned move. , and therefore the dataset move analysis assumes that the move has been performed. Based on the assumption that

Intersubsystem cache reorganization The inter-subsystem cache reorganization routine 405 handles the cache reorganization routine 405. Exists to support movement of cache candidate datasets. cash bo The volume generation function is configured to identify where underutilized cache exists and also to Cache candidate data that is not cacheable due to lack of cache space on the system. Determine where the tasset is located. DASD performance optimization mechanism module 4 02 then looks for underutilized caches and uses these cache subsystems. Cache candidate data that is flagged for movement as being moveable to Trying to find the tasset. If the DASD performance optimization mechanism 402 There is a lot of good cache candidate data that has not been cached, and Check that none of the configured subsystems has sufficient cache capacity. If so, then this module will allow the user to cache some extra Write out a message to the user advising them to obtain it.

identification The job of the identification module 406 is to find overutilized subsystems and volumes. Is Rukoto. A volume or subsystem is a volume or subsystem whose amount of activity is When the amount of queues on a device or route reaches an unacceptable level, the It becomes a problem. Volumes or subsystems are Identification of overuse when activity levels are reached below a difficult level be done. This safety margin allows you to detect potential problem areas. Consider suppressing actual problems as well as preventing problems from occurring. ing.

Detection of these overutilizations is done by checking the busy rate (device) or route (bus). (percent busy) and the number of routes. The busy rate is measured in seconds. I10 number per unit, average data size transferred, channel speed, device Derived from service time and number of routes. which devices and subsystems As well as identifying overutilization, this module also determines how much data needs to be moved from the device or subsystem. this is the route time per second or devices per second that we need to reduce expressed by the amount of time.

rules Possible used to determine if a device or subsystem is overutilized Some examples of the types of rules with gender are shown below, where a and b are approximately It is a predetermined variable.

If the subsystem is not a cache subsystem, and it is a subsystem of an 8380 class device, and There are two paths, and There is no actuator level buffering (ALB), and route is greater than a busy, then If the subsystem is overutilized then the device is in the non-cached subsystem. is on the stem, and It is an 8380 class device and the device is greater than bbusy. , then the device is overused. If the subsystem is a cache subsystem and There are two paths, and the device is an 8380 class device, and ALB does not exist, and If route is greater than a busy (due to staging and 110 route times), then Then the route is overused Identification module 406 identifies overutilized subsystems and volumes. In addition, calculate the overutilization of each of these overutilized subsystems. example For example, if a subsystem is identified as overutilized, identification module 406 may For example, input and output to the various volumes that make up this subsystem. The activities processed by this subsystem by summing the activities. Determine the amount of tea.

Identification module 406 then identifies subsystems related to extra activities. Determine the capacity of In other words, the subsystem can be processed without becoming overburdened. , determine the number of extra input and output operations. Once this calculation is complete, Identification module 406 assigns a classification attribute to this particular subsystem and identifies its The attribute indicates the degree of overutilization of the subsystem. This procedure is the same as overutilization. for each subsystem within a defined overall system, and likewise for overutilization. This is also done for each volume identified as surplus.

Target selection Once overutilization is identified, target selection routine 407 determines whether the volume or is a combination of datasets that solves a problem when moved out of a subsystem. Find out.

The target selection routine 407 selects three alternating combinations. They are profitable Known as a target for each overused volume or subsystem. each The target is a set of data sets whose movement from the volume corrects overutilization. .

The reason for selecting three targets is to avoid overuse, and if the data set If there is a lack of space to place , then there are three different objects, One of them is movable. The selection of the target is done through the candidate generation routine 411. First generate candidate targets and then select the best three through candidate selection routine 412. It is executed by selecting a candidate.

Candidate generation Candidate generation routine 411 is executed if the volume or subsystem is moved. Sometimes the combined activities are sufficient to correct the overutilization of the data set. Generate candidate combinations of cuts. The candidate generation routine 411 identifies overutilization. For each volume in the conformance class to which the specified volume is assigned, Determine space availability and activity capacity. Availability is the amount of available space. activity capacity is a variable that indicates how much Indicates that it can be used. Of greater importance is the availability of one or the other, rather than Given for the presence of both space and activity capacity on the volume . In addition, the same calculations can be used to transfer datasets from one subsystem to another. is carried out to identify the possibility of moving to the system.

This means that all volumes within a matching class are part of one subsystem within the system. This is the case when there are more. Therefore, the dataset can be divided into different subsystems. However, relocating to keeps the data set within the same fitness class, but Reduce the usage level of the subsystem that originally had the datasets stored there. Make it less. Thresholds are calculated to reduce the number of datasets that have to be considered. data sets with activity below the threshold are ignored. Assemble When generating a fit, immobile datasets are excluded and the combined datasets are The selected activity is chosen near the activity requested for the move. . By choosing a candidate to tune a subsystem problem, the candidate It is assumed that this will also resolve any volume issues that may exist within the system. It will be done.

Candidate selection Candidate selection routine 412 is of interest to a volume or subsystem. Select three candidates. While performing this task, the candidate selection routine 412 calculates the degree of difficulty of placing datasets on other volumes. this The degree of change from current location on a particular overutilized volume to less Additionally, moving the dataset to other candidate volumes in the same fitness class , indicating the expected difficulties. While investigating the placement of datasets, we decided to use candidate selection rules. The check box 412 determines whether the destination volume is compatible with the source volume. decide. In other words, the destination volume is part of the same set of volumes as the source volume. Must be shared by hosts and must be within the same user-defined pool. No. In addition, candidate selection routine 412 selects non-cached data sets from non-cache Attempt to move to cache volume. The data set source volume and Apart from finding a matching volume, the candidate selection routine 412 A volume with enough space for the dataset and relocation of the dataset. A volume with sufficient capacity for the extra activity resulting from You have to decide the time. One of the goals used in the selection process is to The goal is to try to minimize the number of data sets that need to be performed. This is because we strive to minimize the number of movements. Some of the candidate targets are It is still possible to have an immovable dataset in the Another goal of processing is to try to avoid selecting subjects with immobile data sets. It is. Large datasets are excluded and subsystems are excluded. Selecting solves the volume problem within the subsystem. Other goals are three It is to strive to create two mutually different objects, so that one object The fact that one cannot be moved does not necessarily mean that the other cannot be moved either. It becomes.

Overuse treatment The overutilization handling routine 408 finds the highest priority and All overutilizations are resolved by calling the overutilization routine 414 to adjust. Deal with the surplus.

The intent is to avoid overutilization or adjustments that significantly impact the I10 subsystem. Overutilization that is difficult to manage should receive the highest priority and be dealt with first. That's what it means.

Find the best overutilization Overutilization priorities are determined based on the degree of overutilization and the expectations for correcting overutilization. Based on the degree of difficulty faced. How difficult is the amount of space and activity? What is the overutilization given in terms of the availability of capabilities and the targets chosen? Based on what you need. The degree of overutilization depends on the I10 performance of the device or or subsystem. Highest priority overuse To find out, the highest priority overutilization discovery routine 413 determines the degree of overutilization. Compare and evaluate the degree of difficulty.

Correct overutilization Given the overutilization, the overutilization correction routine 414 Should sets of files be moved from a volume or subsystem and where should they be moved? Adjust the problem by deciding whether to move. Which pair is this? target selection to determine whether the object should be moved from the volume or subsystem. This is accomplished by calling selection routine 415. Dataset placement routine 416 then to determine where to place the dataset in the object. used.

Select target The target selection routine 415 selects a device with available space and activity capacity. Select the object to be moved by finding the best object with the data set.

Dataset placement The dataset placement routine 416 determines the best place to place the dataset. do. Data sets residing on shared volumes are shared by the same host. Datasets must be constantly moved only to volumes that are Do not cause the previous volume to become overutilized. What if data Set placement routine 416 ensures that sufficient space is available for the data set. If it finds that the space does not exist, it creates enough space to Consider moving one other dataset.

The dataset placement routine 416 removes datasets from overutilized volumes or or subsystem to an underutilized volume on an underutilized subsystem, Moving. Another part of the process of determining where to move the dataset is data set to create space or sufficient activity capacity. It may be moved to make it available. Transferring the original dataset The move list created to accomplish the move is stored in memory. data center There are two different ways to place the cuts. The first is the same as the dataset. on underutilized subsystems that are in the same compatibility class and to which the dataset fits. This is the case when there is an underutilized volume. In this case, the dataset placement Routine 416 selects the volume within this subsystem with the least activity. Place the dataset in one of the volumes. Datasets can be placed A second possible method is the case where there is no volume that meets the above criteria.

In this case, the dataset placement routine 416 may cause one dataset to be transferred to another. Enough space or activity with respect to the original dataset when it was moved. Identify a pair of volumes of the same type that have the same capacity available.

summary A memory performance improvement device is a memory performance improver that can be used to improve computer system errors that occur when attempting to access data storage devices; Identify memory operation conflicts, such as data storage performance degradation. Improved memory performance The device determines which data sets stored on these data storage devices are involved in this contention. Identify motion conflicts. Once a data set is involved in an operation conflict, Once the data sets have been identified, the memory performance improver can determine alternate memory storage locations and replace these conflicting datasets with alternative data storage locations. Invoking various software routines to navigate to a location. These conflicts Data set relocation resolves memory operation conflicts and Improve searchability of data stored in . Conflict identification and dynamic, real-time By performing base resolution, a computer system's data storage operate in a more efficient manner and scan the data stored on these data storage devices. Searchability is significantly improved without the need for data management personnel. Computer A system memory performance improver is therefore a device connected to a computer system. Monitor and correct data storage performance.

While specific embodiments of the invention have been set forth, those skilled in the art will appreciate that the accompanying claims can and will plan variations on this system that fall within the scope of It is expected that

FIG, 2゜ FIG, 3゜ FIG, raw Procedural amendment (formality) %formula% 16 Incident display PCT/US 89105709 2. Name of the invention Computer system memory performance improvement device 3, person who makes corrections Relationship to the incident: Patent applicant Name Storage Technology Corporation 4. Agent Address: 8-10-5, Toranomon-chome, Minato-ku, Tokyo 105, Date of amendment order → 6. Subject of correction Translation of the description and claims 7. Contents of correction Translation of the specification and claims (no changes to the content) 8. List of attached documents Translation of the description and claims - one copy each of the international search report

Claims (32)

[Claims] 1. Stores information representing the configuration of computer system memory (160, 168) (171) and a set of functional rules representing data management functions. means (108) for; memory operation conflicts in the computer system memory (160, 180); for detecting said configuration (171) and rule storage (108) means; means for responding (201, 202, 211); the computer system to be relocated to resolve the memory operation conflict; to identify a data set in the system memory (160, 180), means (203, 212) for responding to the output means (201, 202, 211); data in the memory (160, 180) of the computer system (101). dynamically change the placement of data to improve the searchability of the data stored therein. A system for reorganizing. 2. the method for exchanging memory storage locations to resolve the memory operation conflict; for transporting the identified data set, the identification means (203, 21 2) means (204, 213) for responding to; The system of claim 1, further comprising: 3. The detection means (201, 202, 211) read/write activity of the data set in memory (160, 180) Claim 1, comprising means (201, 211) for monitoring the The system included. 4. The detection means (201, 202, 211) Indicates the usage frequency and location of the data set stored in the memory (160, 180). statistics from the read/write activity of the monitored dataset. Claim 3 further comprising means (202) for calculating data. The system. 5. The computer system memory (160, 180) comprises a plurality of DASD devices. (150-0 to 150-15), and the identification means (203, 212) , the most utilized of the DASD devices (150-0 to 150-15) means (407) for describing the and the least utilized; 5. The system of claim 4, comprising: 6. The identification means (203, 212) identifies these DASD devices (1). 50-0 to 150-15) to balance the activity on the device. In order to exchange between sets, the DASD device (150-0 to 150-15 ), and the DASD devices ( 150-0 to 150-15) from the device described as being used the least , for selecting a data set, responsive to said describing means (407); 6. The system according to claim 5, further comprising means (411, 412). 7. Activities on the DASD devices (150-0 to 150-15) described above In order to balance the -15) for the purpose of exchanging said selected datasets between said said data sets; means (204, 213) responsive to the determining means (203, 212); 7. The system of claim 6, further comprising: 8. The computer system memory (160, 180) comprises a plurality of DASD devices. (150-0 to 150-15) and cache memory (160). The identification means (203, 212) Good and bad candidates for relocation to the cache memory (160) data set on one DASD device (150-0 to 150-15) means for classifying all the items (407); 5. The system of claim 4, comprising: 9. The identification means (203, 212) identifies the cache among the data sets. The DASD is classified as a good candidate for relocation to memory (160). Describe the data set that can be stored on one volume in the device (150-0) 9. The system of claim 8, further comprising means (407) for. 10. In order to resolve the memory operation conflict, the described data set is for transporting to the one volume in the DASD device (150-0). further comprising means (416) for responding to the identification means (203, 212). The system according to claim 9. 11. Information representing the configuration of the computer system memory (160, 180) a functional process representing the memory management functions of the computer system; storing a set of rules (108); 60, 180); a step of detecting memory operation conflicts based on the monitoring results; and resolving the memory operation conflicts. The computer system memory (160, 18) to be relocated to determine 0) identifying a dataset in To resolve the identified memory performance conflict, memory storage locations should be rotated. , transporting the recognized data set; data in computer system memory (160, 180) comprising each step of How to improve search efficiency. 12. The step of detecting retrieving data sets in said computer system memory (160, 180); monitoring read/write activity; 12. The method of claim 11, comprising: 13. The step of detecting the data set stored in the computer system memory (160, 180); read/write of the monitored data set representing usage frequency and location of the site; The process of calculating statistical data from the activities of 13. The method of claim 12, further comprising: 14. The computer system memory (160, 180) comprises a plurality of DASD devices. (150-0 to 150-15), and the identifying step comprises: Among the DASD devices (150-0 to 150-15), the one that is used the most and a step of describing the minimum amount to be used; Activities on these listed DASD devices (150-0 to 150-15) The DAS It was described as being the most utilized of the D devices (150-0 to 150-15). from the device, and to the smallest of said DASD devices (150-0 to 150-15). selecting a data set from the devices described as being utilized; 14. The method according to claim 13, comprising the steps of: 15. The identifying step includes: Activities of these listed DASD devices (150-0 to 150-15) In order to keep the balance of the The selected data set is between the most used and the least used. the process of replacing the 15. The method of claim 14, further comprising: 16. The computer system memory (160, 180) comprises a plurality of DASD devices. (150-0 to 150-15) and cache memory (160). The process of identifying the cache memory (160) is suitable for relocation to the cache memory (160). The candidates and bad candidates have data sets on one DASD device (150-0). 14. The method according to claim 13, comprising the step of classifying all the items. 17. The identifying step includes: Among the data sets, which one is suitable for relocation to the cache memory (160)? One of the components in the DASD device (150-0) that is classified as a candidate describing a dataset storable on the 17. The method of claim 16, further comprising: 18. In order to resolve the memory operation conflict, the described data set is transporting to the one volume in the DASD device (150-0); 18. The method of claim 17, further comprising: 19. Computer system memory (160, 180) includes cache memory (1 60) and multiple DASD devices (150-0 to 150-15) represents the configuration of the computer system memory (160, 180); a set of functional means for contracting information (171) and representing data management functions; means (108) for storing rules; The cache memory (160) and the DASD device (150-0 to 150- The purpose is to detect memory operation conflicts in both the configuration (15) and the configuration (17). 1) and means (201, 202, 211) for responding to the rules storage (108) means. )and, the computer system to be relocated to resolve the memory operation conflict; to identify a data set in the system memory (160, 180), means (203, 212) for responding to the output means (201, 202, 211); data in the memory (160, 180) of the computer system (101). dynamically change the placement of data to improve the searchability of the data stored therein. A system for reorganizing. 20. In order to resolve the memory operation conflict, the previous for transporting the identified data set, the identification means (203, 2 12) means (204, 213) for responding to; 20. The system of claim 19, further comprising: 21. The detection means (201, 202, 211) is connected to the computer system. Data set read/write activity in memory (160, 180) means (201, 211) for monitoring the computer system; Indicates the usage frequency and location of the data set stored in the memory (160, 180). statistics from the read/write activity of the monitored dataset. as claimed in claim 19, comprising means (202) for calculating data. The system. 22. The identification means (203, 212) identifies the DASD devices (from 150-0 to 150-15), indicate which ones are used the most and which ones are used the least. means (407) for Activities on these listed DASD devices (150-0 to 150-15) The DAS It was described as being the most utilized of the D devices (150-0 to 150-15). from the device, and to the smallest of said DASD devices (150-0 to 150-15). To select a dataset from the equipment described as being used, and means (411, 412) for responding to the description means (407). 22. The system according to claim 21. 23. Actuators on the DASD devices (150-0 to 150-15) described above In order to balance the 0-15) for exchanging the selected data set between the means (204, 213) responsive to the identification means (203, 212); 23. The system of claim 22, further comprising: 24. The identification means (203, 212) sends information to the cache memory (160). For relocation, one DASD device (150 - 0) means (311) for classifying all of the above datasets; Among the data sets, which one is suitable for relocation to the cache memory (160)? One of the components in the DASD device (150-0) that is classified as a candidate means (312) for describing a data set, storable on the data set; 22. The system of claim 21. 25. In order to resolve the memory operation conflict, the described data set is for transporting to the one volume in the DASD device (150-0). further comprising means (213) responsive to the identification means (203, 212). 25. The system of claim 24, comprising: 26. Computer system memory (160, 180) includes cache memory (1 60) and multiple DASD devices (150-0 to 150-15). represents the configuration of the computer system memory (160, 180); a means for storing information (171) and a set of functions representing data management functions; means (108) for storing rules; retrieving data sets in said computer system memory (160, 180); means (201, 211) for monitoring read/write activity; a data set stored in the computer system memory (160, 180); read/write access for the monitored dataset, representing the usage frequency and location of the data set; means (202) for calculating statistical data from said activity; Detecting memory operation conflicts within computer system memory (160, 180) and responsive to said configuration (171) and rule storage (108) means. means (201, 202, 211); Among the DASD devices (150-0 to 150-15), the one that is used the most and the means (411, 412) for describing the minimum usage of these; Activity on DASD devices (150-0 to 150-15) described in The DASD device (150 -0 to 150-15), from the most used and the least utilized of the DASD devices (150-0 to 150-15). to select a dataset from those described as means (407) responsive to the means and relocation to said cache memory (160); For a good candidate and a bad candidate, the data on one DASD device (150-0) means (311) for classifying all data sets; The previous cache memory (160) is classified as a good candidate for relocation. data that can be stored on one volume in the DASD device (150-0). and means (312) for resolving the memory performance conflict. The computer system memory (160, 180) to be relocated to ), the detection means (201, 202, 211) and means (203, 212) for responding to these DASD devices. In order to balance the activities on the locations (150-0 to 150-15), Among the DASD devices listed (150-0 to 150-15), the selected for exchanging data sets, and in response to the identification means (203, 212); means (414, 415) to To resolve the memory operation conflict, the described data set is transferred to the DAS. In order to transport the above-mentioned one volume in the D device (150-0), the above-mentioned means (416) responsive to the identification means (203, 212); from the DASD device (150-0) to the cache memory (160). means for writing (405); and in the memory (160, 180) of the computer system (101). Dynamic data placement to improve the searchability of the data stored there. A system for reorganizing. 27. Computer system memory (160, 180) includes cache memory (1 60) and multiple DASD devices (150-0 to 150-15) represents the configuration of the computer system memory (160, 180); a process of recording information; A program for storing a set of functional rules representing the computer system memory management functions. With mode, The cache memory (160) and the DASD device (150-0 to 150- 15) monitoring the operation of both; a step of detecting memory operation conflicts based on the monitoring results; and resolving the memory operation conflicts. The computer system memory (160, 18) to be relocated to determine 0) identifying a dataset in the identified data set to resolve the identified memory operation conflict; transporting the data to an alternate memory storage location; each step of the computer system memory (160, 180). How to improve data search efficiency. 28. The detection step includes: retrieving data sets in said computer system memory (160, 180); monitoring read/write activity; a data set stored in said computer system memory (160, 180); read/write activity of the monitored data set representing usage frequency and location of a process of calculating statistical data from the activity; 28. The method according to claim 27, comprising the steps of: 29. The identification step Among the DASD devices (150-0 to 150-15), the one that is used the most and a step of describing the minimum amount to be used; Activities on these listed DASD devices (150-0 through 150-15) The DASD device is used for exchanging between data sets in order to balance the data sets. (150-0 to 150-15) listed as being used the most and the least available of said DASD devices (150-0 to 150-15). selecting a dataset from those used; 28. The method according to claim 27, comprising the steps of: 30. The identification step Good and bad candidates for relocation to the cache memory (160) , classifying all the data sets on the DASD device (150-0); Among the data sets, which one is suitable for relocation to the cache memory (160)? One of the components in the DASD device (150-0) that is classified as a candidate a step of describing what can be stored on the top; 28. The method according to claim 27, comprising the steps of: 31. within the DASD device (150-0) to resolve the memory operation conflict. transporting the described data set to the one volume of 31. The method of claim 30, further comprising: 32. Computer system memory (160, 180) includes cache memory (1 60) and multiple DASD devices (150-0 to 150-15) represents the configuration of the computer system memory (160, 180); a process of recording information; storing a set of functional rules representing computer system memory management functions and, The cache memory (160) and the DASD device (150-0 to 150- 15) monitoring the operation of both; retrieving data sets in said computer system memory (160, 180); the process of monitoring read/write activity and the computer system menu; Indicates the usage frequency and location of datasets stored in Mori (160, 180), Statistical data from the read/write activity of the monitored dataset and detecting memory operation conflicts from the monitoring results. process and maximum utilization of the DASD devices (150-0 to 150-15). The process of describing what is used and what is minimally used, and the description of these Balance of activity on DASD devices (150-0 to 150-15) The DASD device (150-0 150-15), and from the most utilized one, and the DASD From the least utilized devices (150-0 to 150-15), data For the process of selecting a data set and relocating it to the cache memory (160). The entire data set on the DASD device (150-0) for good and bad candidates. of the data set to the cache memory (160); In the DASD device (150-0) that is classified as a good candidate for relocation. The above-mentioned monitoring method includes each step of describing what can be stored on one volume. a step of detecting memory operation conflicts from the visual results; Activities on these listed DASD devices (150-0 to 150-15) In order to balance the -15) exchanging said selected data set between said memory operations; In order to resolve the conflict, the one policy in the DASD device (150-0) transporting the described data set to the system; data retrieval of computer system memory (160, 180); How to improve efficiency.