International Business Machines Corporation

Yosef Moatti

Erez Hadad

Ofer Biran

Elliot Karl Kolodner

Paula Kim Ta-Shma

David Breitgand

Dean Lorenz

Evgeny Hazanovich

William Frank Micka

Michael Edward Factor

Richard Edwin Harper

Dmitry Sotnikov

Yoram Novick

Alain Charles Azagury

Kenneth Nagin

Oded Cohn

Yaron Weinsberg

Sadek Jbara

Amir Epstein

Lorrie Ann Tomek

George Goldberg

Ari Freund

Guy Khazma

Gil Rapaport

Guy Gerson-Golan Guy Gerson-Golan

Elliot K Kolodmer

Eyal Zangi

Zvi Rosberg

Moshe Bach

David Allon

Abraham Teperman

Efaim Zeidner

A hub of a computing environment obtains a training set from an edge of the computing environment. The training set that is obtained includes data from the edge and is used to train a neural network at the hub. The training of the neural network provides a detector and a generator at the hub. A determination is made as to whether the training of the neural network is complete. Based on determining that the training of the neural network is complete, the detector is sent to the edge. The detector at the edge is to facilitate suppression of additional edge data to the hub based on the detector at the edge determining that the additional edge data is statistically similar, based on one or more selected criteria, to data used to train the neural network.

Reduction of data transmission and data storage using neural network technology

Embodiments may include novel techniques to communicate user preferences to the FaaS provider so as to provide full applicability of FaaS for business critical applications and to provide full realization of the FaaS model flexibility. For example, in an embodiment, a method may be implemented in a computer system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, and the method may comprise receiving a request for processing of a computing task and associated data indicating a context of an overall process flow for the computing task, wherein the request for processing is a function invocation for a FaaS computing system, evaluating the data indicating the context and scheduling computing resources for performing the computing task based on the data indicating the context, and performing the computing task using the scheduled computing resources.

System and apparatus for FaaS business goals optimization

Improved computer data query processing by a) identifying a cut point value (CPV) for a data set, where an attribute value of each data set record is normalized between first and second normal bounds, where the CPV is any value between the normal bounds that is associated with a maximum data skipping performance value (DSP) calculated by applying a DSP function based on the normalized values and the CPV, and where the DSP function relates to a selected data query distribution profile, b) partitioning the data set into multiple partitions, one partition including a data record subset whose corresponding normalized attribute values are greater than the CPV, and another partition including a data record subset whose corresponding normalized attribute values are less than the CPV, and c) storing the partitions on physical data storage media for access by a computer configured to perform data query processing.

Data partitioning for improved computer performance when processing

Systems and methods for allocating resources to service one or more requests, the method comprising: associating with at least one host from among N hosts, in a computing environment, a first value that reflects a probability that servicing a first request will interfere with operations on the host; selecting M hosts out of the N hosts, wherein the N hosts comprise one or more resources for servicing a first request and wherein at least one of the selected M hosts is determined to be at least as suitable as N&#x2212;M unselected hosts, for the purpose of servicing the first request; and prior to allocation of resources, updating for at least a first host from among the M hosts the corresponding first value to reflect that servicing the first request may interfere with operations pertaining to other previously received requests that are scheduled to be serviced by the first host.

Efficient resource management in a virtualized computing environment

Methods, computing systems and computer program products implement embodiments of the present invention that include defining a first multiple of software container configurations and a second multiple of permission sets, and receiving, by a first computer, a request to perform a service operation on a second computer having multiple resources. Upon identifying one or more of the resources that are required for the service operation, a given software container configuration and a given permission set are selected based on the identified one or more resources, and the given software container configuration and the given permission set are conveyed to the second computer. Upon the second computer receiving the given software container configuration and the given permission set, a software container is generated. The software container is opened on the host computer prior to performing the service operation, and closed upon completing the service operation.

Container-based system administration

Machines, systems and methods for managing resources allocated in a virtualized computing network are provided. The method comprises monitoring resource usage by a first host, wherein the first host, in addition to the first host's dedicated resources, has access to a first set of sharable resources, and wherein the second host, in addition to the second host's dedicated resources, has access to a second set of sharable resources; determining whether the second host has excess resource capacity in the second host's dedicated and sharable resources, in response to determining that resource usage of the first host in the first host's dedicated and sharable resources has exceeded a resource usage threshold; and reallocating one or more resources in the second set of sharable resources from the second host to the first host, in response to determining that the second host has excess resource capacity.

Optimized global capacity management in a virtualized computing environment

Machines, systems and methods for deciding whether a plurality of virtual machines (VMs) may be hosted on a computing system with resources to support at least one of the VMs, the method comprising determining maximum permissible probability of infringement (maxPermInfrProb(v)) for a VM(v), wherein 'v' is a value from 1 to n, such that n is the number of VMs in a set G, as stipulated by the VMs' SLA, wherein 'v' is a value from 1 to n, such that n is the number of VMs in a set G; partitioning the set G into a plurality of groups G(G, . . . G, . . . G), so that each group is homogenous with respect to maxPermInfrProb defined for members of the group G; and organizing the groups Gbased on the value of maxPermInfrProb for each G.

Resource allocation in a virtualized computing environment

Systems and methods for planning the execution of one or more plans in a virtualized computing environment are provided. In one embodiment, the method comprises computing an order for execution for one or more plans, and the corresponding operations in said one or more plans; validating whether a plan and the corresponding operations for the plan may be successfully executed, prior to executing the plan and the corresponding operations; executing the plan and the corresponding operations, in response to a successful validation; and aborting execution of the plan and the corresponding operations, in response to an unsuccessful validation, wherein the validating process takes into consideration impact of other plans that are previously validated and possible changes in the computing environment since time the plan was computed.

Concurrent scheduling of plan operations in a virtualized computing environment

Methods and systems for generating a placement plan for one or more virtual machines (VMs) in a computing environment are provided. The method comprises providing, to a computing system, input parameters comprising a current placement for the one or more VMs on one or more hosts in a computing network; a target placement that assigns at least one of the one or more VMs to at least another host in said one or more hosts; and a set of constraints with which both the current placement and the target placement comply.

Optimized placement planning for virtual machines in a network by breaking relocation graph into one or more cores to apply different solutions

Systems and methods for reducing risk of service interruptions for one or more virtual machines (VMs) in a computing environment are provided. The method comprises computing a placement scheme for placing at least one VM on one or more hosts according to a set of placement constraints defined for the VM, wherein the set of placement constraints comprises at least one availability constraint defined for the VM, wherein the availability constraint designates a N resiliency level, wherein N corresponds to number of host failures that may occur before the services provided by the VM are interrupted.

Optimized placement of virtual machines in a network environment

Systems and methods for domain management in a virtualized computing environment are provided. In one embodiment, the method comprises collating advice received from one or more domain advisors connected in the virtualized computing environment; resolving any conflicts among the advice received from said one or more domain advisors; utilizing the collated advice to generate a placement plan comprising a plurality of operations for virtual machines in said virtualized computing environment; and executing the one or more operations in the placement plan, wherein one or more domain handlers may be called to update the virtualized computing environment before, during or after execution of one or more operations from among said plurality of operations in the plan.

Domain management and integration in a virtualized computing environment

A system and method for planning placement of virtual machines VMs in a computing environment comprising a set of hosting machines HM. The method includes constructing a bipartite directed graph-based model that represents both a current and a target placement states, both including virtual machine nodes v in VM and nodes h in HM. Using a technique of graph pruning, the method iteratively generates a plan for achieving a desired target placement starting from the current placement without temporarily violating any policy or resource constraint. The application of the pruning algorithm to VM deployment automation necessarily defines a new model. The methodology employed safely manages concurrent changes in a datacenter, so that the environment can adjust faster to changing constraints. Furthermore, the present invention allows detection of migration cycles, caused by deadlocked combinations of capacity and policy constraints, and resolving them.

System and method for deploying virtual machines in a computing environment

A method for performing remote calls between source and target computing machines includes running a program thread on the source machine which invokes a remote call to the target machine. The remote call is transmitted to the target machine, the call including an identifier associated with the program thread. A response to the remote call is received from the target machine, the response including the identifier, whereby the response is returned to the program thread on the source machine using the identifier.

Preserving thread identity during remote calls

A method and apparatus for improving storage controller throughput and reducing unit control block contention when selecting a Utility Volume for servicing non-specific read requests. Throughput is increased by enabling the storage controller to service non-specific read requests issued against a Utility Volume which is not the same as the device associated with the data that the non-specific read is requesting. The Utility Volume is dynamically selected by considering the current loads on the available devices for both specific and non-specific processing. A Penalty Index is assigned to each volume, wherein the Penalty Index is a weighted average of the specific and non-specific processing. The device having the lowest Penalty Index is selected as the Utility Volume for a session. A Penalty Index evaluation is made at a Reference Point which is further enabled by a random clock to ensure that sessions do not likely pick the same Utility Volume.

Growing community of inventors

Haifa, Israel

Yosef Moatti