International Business Machines Corporation

Chetan L Gaonkar

Vamshi Krishna Thatikonda

Suma M B Bhat

Keerthi B Kumar

Chidambar Y Kulkarni

Lakshmi Priya

Uma Maheswara R Chandolu

Amit Agarwal

Shailaja Mallya

Vishal R Mansur

Vamshikrishna Thatikonda

Anasuya M Shankar

Geeta Devi Akoijam

Niranjan Behera

Sanchita Sinha

Deepak L Ranganath

Balaji B

Prajith P Kumaran

Frederic Durairaj Maria Joseph

Provided is a method, computer program product, and coherent computer system for improving memory management by establishing cooperation between an operating system and a coherent accelerator device (CAD). The CAD may retrieve a set of work elements for completion from a work queue. The CAD may determine a length of time required to complete the set of work elements. The CAD may identify a set of memory pages needed for completing the set of work elements. The CAD may communicate the set of memory pages and the length of time required to complete the set of work elements to a virtual memory manager.

Management of memory pages for a set of non-consecutive work elements in work queue designated by a sliding window for execution on a coherent accelerator

Provided are embodiments of a method for performing live updates. The method includes determining a migration type for a live migration to be performed from a source server to a target server, and performing the live migration from the source server to the target server. The method also includes based on the migration type, maintaining pages to be migrated, tracking modifications in a hypervisor page table, and performing a back-migration based on the migration type. Also provided are embodiments for a system and computer program product for performing live updates.

Increasing the performance of cross the frame live updates

A storage management unit is configured to identify a plurality of page frames in primary storage. The storage-management unit is further configured to determine an amount of unused capacity of each page frame of the plurality of page frames. The storage-management unit is further configured to sort the plurality of page frames based on the determined amount of unused capacity of each page frame.

Management of paging in compressed storage

Disclosed aspects relate to rearrangement management for a shared pool of configurable computing resources including a source asset, a target asset, and a memory device having a set of data. A first subset of the set of data on the memory device may be locked with respect to modification. A first set of logical address identifiers for the first subset of the set of data on the memory device may be identified in a hypervisor mapping table in a first set of source entries for the source asset. The first set of logical address identifiers for the first subset of the set of data on the memory device may be established in the hypervisor mapping table in a first set of target entries for the target asset. The first subset of the set of data on the memory device may be unlocked with respect to modification.

Rearrangement management for a shared pool of configurable computing resources

A page compression strategy classifies uncompressed pages selected for compression. Similarly classified pages are compressed and bound into a single logical page. For logical pages having pages with more than one classification, a weighting factor is determined for the logical page.

Page compression strategy for improved page out process

A physical address of a page may be identified. A first process that implements copy-on-read techniques for the page may be detected. A determination may be made that the first process is not expected to write to the page. In response to that determination, a different logical address may be established for the first process for the page from the logical address of a second process for the page, but the two logical addresses may be mapped to the same physical page.

Implicit sharing in storage management

As disclosed herein a method, executed by a computer, for enabling live partition mobility using ordered memory migration includes receiving a request to initialize a migration of a logical partition (LPAR) to a destination system. The method further includes creating a list which includes memory page identifiers corresponding to memory pages of the LPAR. The memory page identifiers of the list are ordered according to a page transfer priority. The method further includes identifying memory pages of the LPAR that will be unmodified during an estimated duration of time of the migration. The method further includes updating the list, based on the identified memory pages of the LPAR that will be unmodified during the estimated duration of time of the migration. The method further includes migrating the LPAR based on the list. A computer system, and a computer program product corresponding to the method are also disclosed herein.

Live partition mobility using ordered memory migration

A rule-based method for pre-fetching 'important pages' from memory paging space back into kernel memory space performs the following steps: (i) establishing a first machine logic based rule, with the machine logic based rule including a first triggering condition and a first consequential responsive action; (ii) determining that the first triggering condition has occurred; and (iii) in response to the determination that the first triggering condition has occurred, performing the first consequential responsive action. The first triggering condition includes the following sub-conditions: (i) at least one of a first set of important page(s) of a computer system has been paged out of kernel memory space and into paging memory space, and (ii) a processing status of the computer system indicates that the first set of important page(s) will be likely to be required for computer operations.

Virtual machine manager initiated page-in of kernel pages

Embodiments of the present invention provide hints for page stealing by prioritizing pages based on the number of residences. Receiving a plurality of pages to be hinted to a hypervisor for page stealing. Determining at least two page types of the plurality of pages. Determining whether any of the at least two page types has a total number of residences less than a total number of potential residences in the virtual environment for all page types and have a total number of residences less than a threshold. Responsive to determining a first page type of the at least two page types has a total number of residences less than a total number of potential residences for all page types and has a total number of residences less than a threshold, notifying the hypervisor of at least one page from the plurality of pages that is the determined first page type.

Growing community of inventors

Kumta, India

Chetan L Gaonkar