International Business Machines Corporation

Adam Thomas Stallman

Narsimha Reddy Challa

Larry William Youngren

Ryan Kirk Cradick

Daniel Beuch

Michael J Branson

Charles Scott Graham

Swaroop Jayanthi

Sailaja R Keshireddy

Gary Ross Ricard

Chad Allen Olstad

Steven Joseph Branda

Suresh Guduru

Christopher James Kundinger

Brett Michael Yokom

Michael K Branson

An example computer-implemented method for mirroring memory in a disaggregated memory clustered environment is provided. The method includes assigning, by a hypervisor, a disaggregated memory to a virtual machine comprising a remote disaggregated memory, the virtual machine being one node of a cluster of the disaggregated memory clustered environment. The method further includes allocating, by a disaggregated memory manager, a mirrored memory for the remote disaggregated memory to mirror the remote disaggregated memory on an alternate node of the cluster of the disaggregated memory clustered environment. The method further includes responsive to a memory access occurring, maintaining, by the disaggregated memory manager, the mirrored memory. The method further includes, responsive to detecting a memory allocation adjustment, modifying, by the disaggregated memory manager, memory usage across the cluster. The method further includes, responsive to detecting a failure resulting in loss of access to the remote disaggregated memory, implementing a corrective action.

Mirrored disaggregated memory in a clustered environment

A stream of tuples is processed by a stream application. The stream application includes a plurality of processing elements that operate on one or more compute nodes, each processing element includes one or more stream operators. One or more databases that are capable of communicating with the stream application are monitored during the processing of the stream of tuples. A potential performance condition of a first database of the one or more databases is detected based on the monitoring. An output adjustment is performed, in response to the potential performance condition.

Datastore aware streams computing

An embodiment includes analyzing data associated with an original flow graph comprising a plurality of operators of a stream computing application, including identifying a secure network connection between a first operator and a second operator that uses encryption. The embodiment fuses the first operator with the second operator such that a first logical function of the first operator is combined with a second logical function of the second operator. The embodiment then generates a modified flow graph as a modification of the original flow graph that combines the first operator and the second operator and lacks encryption between the first operator and the second operator.

Using fusion to reduce encryption in streaming applications

An embodiment includes monitoring a distributed computing application at runtime for occurrence of a condition, where the condition includes occurrence of a degradation condition on a processing element (PE). The embodiment also includes starting a provisional PE on a second node at runtime while maintaining runtime operation of the PE. The embodiment compares operation of the PE to operation of the provisional PE based on a performance metric associated with the computing resource. The embodiment connects the provisional PE to replace the stream of tuples from the PE to a downstream PE with a stream of tuples from the provisional PE to the downstream PE.

Dynamic replacement of degrading processing elements in streaming applications

Embodiments of the invention provide for methods for the management of logically partitioned computing resources of a data processing system configured with a plurality of hypervisors that each manages one or more logical partitions of the computing resources. Each hypervisor and logical partitions managed thereby are allocated to a domain of a plurality of domains. Logical partitions and hypervisors allocated to a first domain are monitored to determine whether additional computing resources are needed for the first domain. Responsive to determining that additional computing resources are needed for the first domain, a second domain having available computing resources is determined. At least a portion of the available computing resources of the second domain are used by the first domain.

Dynamic management of computing platform resources

Embodiments of the invention provide for systems and computer program products for the management of logically partitioned computing resources of a data processing system configured with a plurality of hypervisors that each manages one or more logical partitions of the computing resources. Each hypervisor and logical partitions managed thereby are allocated to a domain of a plurality of domains. Logical partitions and hypervisors allocated to a first domain are monitored to determine whether additional computing resources are needed for the first domain. Responsive to determining that additional computing resources are needed for the first domain, a second domain having available computing resources is determined. At least a portion of the available computing resources of the second domain are used by the first domain.

Embodiments of the invention provide for systems and computer program products for the management of logically partitioned computing resources of a data processing system configured with a plurality of hypervisors that each manages one or more logical partitions of the computing resources. A plurality of domains for the data processing system may be determined. For each domain, one or more hypervisors may be allocated to the domain such that one or more logical partitions managed by the hypervisor are allocated to the domain. Usage of the logically partitioned computing resources is based at least in part on the domain of each logically partitioned computing resource, a domain of each hypervisor, and/or a domain of a user.

Embodiments of the invention provide for methods for the management of logically partitioned computing resources of a data processing system configured with a plurality of hypervisors that each manages one or more logical partitions of the computing resources. A plurality of domains for the data processing system may be determined. For each domain, one or more hypervisors may be allocated to the domain such that one or more logical partitions managed by the hypervisor are allocated to the domain. Usage of the logically partitioned computing resources is based at least in part on the domain of each logically partitioned computing resource, a domain of each hypervisor, and/or a domain of a user.

Embodiments of the invention provide for systems and computer program products for management of logically partitioned computing resources of a data processing system configured with hypervisors that each manages one or more logical partitions of the computing resources. The data processing system is configured with domains, and allocated to each domain are one or more hypervisors and logical partitions managed thereby. For a first domain, a token is received from a second domain. The token identifies available computing resources of the second domain that are available for temporary allocation. The data processing system determines whether the available computing resources of the second domain are needed by the first domain based at least in part on the received token. Responsive to determining that the available computing resources are needed by the first domain, the available computing resources are temporarily allocated to the first domain for use thereby.

Embodiments of the invention provide for methods for management of logically partitioned computing resources of a data processing system configured with hypervisors that each manages one or more logical partitions of the computing resources. The data processing system is configured with domains, and allocated to each domain are one or more hypervisors and logical partitions managed thereby. For a first domain, a token is received from a second domain. The token identifies available computing resources of the second domain that are available for temporary allocation. The data processing system determines whether the available computing resources of the second domain are needed by the first domain based at least in part on the received token. Responsive to determining that the available computing resources are needed by the first domain, the available computing resources are temporarily allocated to the first domain for use thereby.

A method utilizes priority-based bandwidth allocation to reallocate unused bandwidth capacity of a virtualized IO resource such as a self-virtualizing IO resource to one or more virtual functions in an allocation pool based upon priorities assigned to such virtual functions.

Dynamic sharing of unused bandwidth capacity of virtualized input/output adapters

An apparatus and program product utilize priority-based bandwidth allocation to reallocate unused bandwidth capacity of a virtualized IO resource such as a self-virtualizing IO resource to one or more virtual functions in an allocation pool based upon priorities assigned to such virtual functions.

A method, system and computer program product are provided for implementing dynamic adjustment of Input/Output bandwidth for Virtual Machines of a Single Root Input/Output Virtualization (SRIOV) adapter. The SRIOV adapter includes a plurality of virtual functions (VFs). Each individual virtual function (VF) is enabled to be explicitly assigned to a Virtual Machine (VM); and each of a plurality of VF teams is created with one or more VFs and is assigned to a VM. Each VF team is enabled to be dynamically resizable for dynamic adjustment of Input/Output bandwidth.

Growing community of inventors

Rochester, MN, United States of America

Adam Thomas Stallman