International Business Machines Corporation

Shashidhar Bomma

Neeraj Kumar Kashyap

Pramod V Gavali

Andrew Dunshea

Ginni Gidwani

Sudheesh S Kairali

Douglas J Griffith

Brian W Hart

Joon Chang

Sreenivas Makineedi

Akash V Gunjal

Jean-Philippe Sugarbroad

Ramesh Veeramalla

An approach is provided in which a microservice fabric controller initiates a call flow that includes a set of transactions through a microservices chain that includes a set of microservices. The microservice fabric controller detects a transaction failure at one of the microservices that is positioned in the microservices chain subsequent to at least one successful microservice. In turn, the microservice fabric controller retries the transaction on the failed microservice without disrupting the previously successful microservice.

Achieving atomicity in a chain of microservices

Workload redistribution is provided. Upper and lower thresholds of resource utilization by worker nodes in a worker node group are defined. Hot and cold regions are determined in the worker node group to provide policy-based rescheduling of the worker nodes and redistribution of workload on the worker nodes based on the upper and lower thresholds of resource utilization. Resource utilization data corresponding to the worker nodes in the worker node group are collected at defined time intervals. It is determined whether an average resource utilization of any worker node in the worker node group is greater than the upper threshold of resource utilization based on the collected resource utilization data. In response to determining that the average resource utilization of any worker node in the worker node group is greater than the upper threshold, redistribution of the workload on the worker nodes is triggered.

Redistributing workloads across worker nodes based on policy

A method, computer program product, and system includes a processor(s) progressively recording data modifications to an object (e.g., a virtual resource or a container), in an in-memory resource of the shared computing environment. Based on receiving an indication of a system failure or a system reboot, the processor(s) writes the data modifications to a non-volatile storage resource, where the non-volatile storage resource is readable by an object manager communicatively coupled to the non-volatile storage resource, and where the object manager utilizes the data modifications to recover the object at reboot following the system failure.

Modifying a container instance network

Functionality can be implemented in a virtual memory manager (VMM) to allow small pages (e.g., 4 KB) to be coalesced into large pages (e.g., 64 KB), so that a single free list can be maintained for the large pages ('maintained pages'). When a process requests a small page, the VMM can associate a maintained page with a memory segment accessible by the process. Then, the maintained page can be divided to form a set of small pages ('fragments'). The fragments can become available pages in a broken page list. The VMM can satisfy the request by allocating one of the fragments in the broken page list. If the process requests additional small pages, the additional requests can be satisfied from the broken page list. When the process terminates, the fragments in the broken page list become a maintained page and can be returned to the free list.

Automatic page promotion and demotion in multiple page size environments

A sync manager invokes a scheduled sync task that includes filter attributes for filtering a plurality of memory pages into a subset of memory pages. The sync manager selects file pointers that correspond to a plurality of files and traverses through the files using the selected file pointers. While traversing, the sync manager extracts file attributes for each of the files and, in turn, filters the files by comparing the filter attributes with the extracted file attributes. The sync manager then extracts memory page identifiers that correspond to the filtered files and synchronizes a subset of memory pages using the extracted memory page identifiers.

Enhanced synchronization framework providing improved sync granularity

Some embodiments comprise a method for selecting data to be transferred to a storage space of virtual memory and include identifying a set of data and determining subsets. Determining subsets may allow for delays before transferring the subsets and allow access to memory of the subsets during the delays. Accesses during the delays may enable embodiments to select other data to be transferred to the storage space and prevent transference of the accessed data. Other embodiments comprise apparatuses that have a paging space, a page identifier, and a page transferrer to transfer pages to the paging space after a delay. The delay may prevent a number of pages from being transferred to the paging space, such as for pages that were accessed during the delay.

Data transference to virtual memory

The present invention provides a computer implemented method for truncating a file in a file system. The file system receives a request to truncate a file to a target size, the file having a first inode. The file system creates a second inode to a transient directory, wherein the second inode holds a plurality of in-progress entries. The file system sets a target size in the first inode. The file system writes a truncate entry to a journal, the truncate entry comprising a predetermined size. The file system truncates the file by up to the predetermined size. The file system determines that the file is above the target size. Responsive to a determination that the file is above the target size, the file system repeats the truncating step and the determining step. Responsive to a determination that the file is not above the target size, the file system applies the in-progress entries from the transient directory.

Growing community of inventors

Austin, TX, United States of America

Shashidhar Bomma