International Business Machines Corporation

Yaoping Ruan

Valentina Salapura

Ruchi Mahindru

John Alan Bivens

Min Li

Eugen Schenfeld

Harigovind Venkatraj Ramasamy

Koushik Kumar Das

Sambit Sahu

Milton H Hernandez

Anees Abdul Shaikh

Hai Huang

Constantin Mircea Adam

Zon-Yin Shae

Hani Talal Jamjoom

Sai Zeng

Mark Edward Podlaseck

Anshul Sheopuri

Huiming Qu

Yu Gu

Debasisha Padhi

Frederick Y-f Wu

Jing Min Xu

Shang Qing Guo

Hiroki Nakano

Jim Alain Laredo

Debanjan Saha

Takayuki Kushida

Yuji Sugiyama

Kumar Bhaskaran

Pei Ni Liu

Paul Driscoll

Larisa Shwartz

Hui Lei

Xiao Xi Liu

Mattias Marder

Lawrence David Bergman

John Joseph Rofrano

Raymond Byars Jennings, Iii

Anuradha Bhamidipaty

Nikolaos Anerousis

Frederick Y Wu

Shu Tao

Robert Alan Flavin

Chee Meng Low

Ramendra K Sahoo

Winnie W Cheng

Fabio Minoru Tanada

Ravi B Konuru

Nandagopal Seshagiri

Daniel Dean

Byungchul Tak

Vysakh K Chandran

Jayan Nallacherry

Paul Jennings

Maja Vukovic

Ashish Kundu

Richard Harold Boivie

Jian Min Jiang

Vugranam Chakravarthy Sreedhar

Rajeev Puri

Frederick Yung-Fung Wu

Kunwadee Sripanidkulchai

Michael C Smith

Dingcheng Li

Cashchakanith Venugopal

Denis R Saure

Theophilus Benson

Bradley W Crawford

Kai-Yuan Hou

Supreet R Mandala

Monther Abdullah Al-Dawsari

Checed A Rodgers

Denis Roland Saure

Chang-Hao Tsai

Debanjan Saba

Dingcheng Lil

Creating a summary of a plurality of texts includes tokenizing each of a plurality of texts to obtain tokens; generating a vector space using a first set of vectors having one or more obtained feature scores equal to or larger than a predefined value; executing non-hierarchical clustering using the vector space to generate a first plurality of clusters; choosing a first representative text in each of the plurality of clusters; generating a second set of vectors from each of the arrays generated based on a number of characters included in tokens of the representative texts; executing hierarchical clustering using the second set of vectors to generate a second plurality of clusters; and in response to a determining a number of clusters included in the second plurality of clusters, determining a second representative text for each of the clusters included in the second plurality of clusters.

Creation of a summary for a plurality of texts

Performing an operation comprising transforming an input dataset to a predefined format, extracting, from the transformed dataset, a plurality of features describing the transformed dataset, and generating, by a machine learning (ML) algorithm executing on a processor and based on an ML model, a plurality of rules for modifying the transformed dataset to conform with a first data model.

Cognitive data discovery and mapping for data onboarding

An actuator to execute on a server may be automatically selected based on risk of failure and damage to the server. Requirement specification and environment parameters may be received. A subset of actuators may be selected based on a risk threshold from an actuator catalog database storing actuator information and actuator risk metadata associated with a plurality of actuators. The actuator risk metadata may be augmented with risk information. A ranked list of the subset of actuators may be generated based on the actuator risk metadata associated with each actuator in the subset. An actuator in the ranked list may be executed on the server.

Action recommendation to reduce server management errors

Various embodiments for optimizing memory bandwidth in a disaggregated computing system, by a processor device, are provided. Respective memory devices are assigned to respective processor devices in the disaggregated computing system, the disaggregated computing system having at least a pool of the memory devices and a pool of the processor devices. An analytic function is performed on data resident in the pool of the memory devices using memory bandwidth not currently committed to a primary compute task.

Opportunistic data analytics using memory bandwidth in disaggregated computing systems

Respective memory devices are assigned to respective processor devices in a disaggregated computing system, the disaggregated computing system having at least a pool of the memory devices and a pool of the processor devices. An iterative learning algorithm is used to define data boundaries of a dataset for performing an analytic function on the dataset simultaneous to a primary compute task, unrelated to the analytic function, being performed on the dataset in the pool of memory devices using memory bandwidth not currently committed to the primary compute task, thereby efficiently employing the unused memory bandwidth to prevent underutilization of the pool of memory devices.

Server resources in a data center are disaggregated into shared server resource pools, including an accelerator (e.g., FPGA) pool. Servers are constructed dynamically, on-demand and based on workload requirements, by allocating from these resource pools. According to this disclosure, accelerator utilization in the data center is managed proactively by assigning accelerators to workloads in a fine granularity and agile way, and de-provisioning them when no longer needed. In this manner, the approach is especially advantageous to automatically provision accelerators for data analytic workloads. The approach thus provides for a 'micro-service' enabling data analytic workloads to automatically and transparently use FPGA resources without providing (e.g., to the data center customer) the underlying provisioning details. Preferably, the approach dynamically determines the number and the type of FPGAs to use, and then during runtime auto-scales the FPGAs based on workload.

Utilizing accelerators to accelerate data analytic workloads in disaggregated systems

Embodiments generally relate transforming data for a target schema. In some embodiments, a method includes receiving input data, where the input data includes a plurality of segments, and where the segments include a plurality of source fields containing target data. The method further includes characterizing the input data based at least in part on a plurality of predetermined metrics, where the predetermined metrics determine a structure of the input data. The method further includes mapping the target data in the source fields of the segments to a plurality of target fields of a target schema based at least in part on the characterizing. The method further includes populating the target fields of the target schema with the target data from the source fields based at least in part on the mapping.

Transforming data for a target schema

A new approach to resiliency management is provided in a data center wherein servers are constructed dynamically, on-demand and based on workload requirements and a tenant's resiliency requirements by allocating resources from these pools. In this approach, a set of functionally-equivalent 'interchangeable compute units' (ICUs) are composed of resources from resource pools that have been extended to include not only different resource types (CPU, memory, accelerators), but also resources of different specifications (specs) and flavors. As a workload is being processed, the health or status of the resources are monitored. Upon a performance issue or failure event, a resiliency manager can swap out a current ICU and replace it with a functionally-equivalent ICU. Preferably, individual ICUs are hosted on one of: resources of a same type each with different specifications, and resources of a same type and specification and different flavors. The approach enables failure independence in a disaggregated environment.

Building a highly-resilient system with failure independence in a disaggregated compute environment

For power management in a disaggregated computing system, upon detecting an electrical power reduction to the disaggregated computing system, respective workloads being performed by respective processors within a set of processors are prioritized according to a service level agreement (SLA) of the respective workloads. Backup electrical power is dynamically allocated to the respective processors performing the respective workloads based upon a priority of the SLA of the respective workloads.

SLA-based backup power management during utility power interruption in disaggregated datacenters

Various embodiments for performing hardware upgrades in a disaggregated computing environment. A workload is run on a disaggregated computing system while providing a new component to at least one of a plurality of component pools used by the disaggregated computing system. Point-to-point circuit wire level switching is used to switch the disaggregated system from an assigned component residing in a first of the plurality of component pools to the new component residing in a second of the plurality of component pools without interrupting the running workload.

Live, in-line hardware component upgrades in disaggregated systems

Server resources in a data center are disaggregated into shared server resource pools. Servers are constructed dynamically, on-demand and based on workload requirements, by allocating from these resource pools. A disaggregated compute system of this type keeps track of resources that are available in the shared server resource pools, and it manages those resources based on that information. Each server entity built is assigned with a unique server ID, and each resource that comprises a component thereof is tagged with the identifier. As a workload is processed by the server entity, its composition may change, e.g. by allocating more resources to the server entity, or by de-allocating resources from the server entity. Workload requests are associated with the unique server ID for the server entity. When a workload request is received at a resource, it matches its unique server ID to that of the request before servicing the request.

Specifying a disaggregated compute system

A method is provided for data matching between a set of source data structures and a set of target data structures. The method includes obtaining, using a processor device configured to perform machine learning, source to target matching results with matching scores, based on the sets of source and target data structures. The method further includes calculating, by the processor device, context information for data structure pairs based on a structure similarity and an ontology similarity between constituent data structures thereof. Each of data structure pairs include as the constituent data structures a respective source data structure and a respective target data structure from the sets of source and target data structures. The method also includes updating, by the processor device, the matching scores based on the context information. The method additionally includes controlling, by the processor device, a hardware device responsive to at least one updated matching score.

Data matching accuracy based on context features

A group of processors in a processor pool comprise a secure 'enclave' in which user code is executable and user data is readable solely with the enclave. This is facilitated through the key management scheme described that includes two sets of key-pairs, namely: a processor group key-pair, and a separate user key-pair (typically one per-user, although a user may have multiple such key-pairs). The processor group key-pair is associated with all (or some define subset of) the processors in the group. This key-pair is used to securely communicate a user private key among the processors. The user private key, however, is not transmitted to non-members of the group. Further, preferably the user private key is refreshed periodically or upon any membership change (in the group) to ensure that non-members or ex-members cannot decipher the encrypted user key.

Securely transferring computation in a disaggregated environment using a processor group key

Embodiments for performing rolling software upgrades in a disaggregated computing environment. A rolling upgrade manager is provided for upgrading one or more disaggregated servers. A designated memory area is used for storing an updated software component, and a disaggregated server is switched to the designated memory area from a currently assigned memory area when performing the software upgrade. A process state and program data is maintained in the currently assigned memory area while maintaining the updated software component in the designated memory area such that the process state and program data are read from the currently assigned memory area and the updated software component is read from the designated memory area during currently executing operations of the disaggregated server.

Rolling upgrades in disaggregated systems

Growing community of inventors

White Plains, NY, United States of America

Yaoping Ruan