Artificial intelligence and machine learning infrastructure
AssigneePure Storage, Inc.
InventorsBrian Gold, Emily Watkins, Ivan Jibaja, Igor Ostrovsky, Roy Kim
Life of this patent
Estimating the term…
Abstract
An artificial intelligence and machine learning infrastructure system, including: one or more storage systems comprising, respectively, one or more storage devices; and one or more graphical processing units, wherein the graphical processing units are configured to communicate with the one or more storage systems over a communication fabric; where the one or more storage systems, the one or more graphical processing units, and the communication fabric are implemented within a single chassis.
Claims19
A method of storage system query processing, the method comprising:
- receiving, at a storage system from one or more graphical processing units (GPUs) via a storage system application program interface (API) provided by the storage system directly to the one or more GPUs and configured to provide specification of one or more data storage operations, one or more data storage operations specifying storage of multiple data objects that respectively include data and metadata describing one or more attributes of the data;
- receiving, at the storage system from the one or more GPUs via the storage system API further configured to provide specification of one or more queries that operate on metadata for the multiple data objects, a query that includes metadata that specifies one or more attributes of data;
- generating, based on the storage system searching through the metadata of data objects stored in the storage system, a dataset that includes one or more of the multiple data objects such that the metadata of each data object in the dataset satisfies the one or more attributes of data specified by the metadata included in the query;
- and
- transmitting, from the storage system to the one or more GPUs, the dataset of the one or more of the multiple data objects.
An artificial intelligence and machine learning infrastructure system comprising:
- one or more storage systems comprising, respectively, one or more storage devices included within respective storage arrays controlled by a respective one or more storage controllers;
- and
- one or more graphical processing units, wherein the graphical processing units are configured to communicate with the one or more storage systems over a communication fabric;
- wherein the one or more storage systems, the one or more graphical processing units, and the communication fabric are implemented within a single chassis;
- and
- wherein the one or more storage systems are configured to:
- receive, at a storage system from the one or more graphical processing units via a storage system application program interface (API) provided by the storage system directly to the one or more graphical processing units and configured to provide specification of one or more data storage operations, one or more data storage operations specifying storage of multiple data objects that respectively include data and metadata describing one or more attributes of the data;
- receive, at the storage system from the one or more graphical processing units via the storage system API further configured to provide specification of one or more queries that operate on metadata for the multiple data objects, a query that includes metadata that specifies one or more attributes of data;
- generate, based on the storage system searching through the metadata of data objects stored in the storage system, a dataset that includes one or more of the multiple data objects such that the metadata of each data object in the dataset satisfies the one or more attributes of data specified by the metadata included in the query;
- and
- transmit, from the storage system to the one or more graphical processing units over the communication fabric, the dataset of the one or more of the multiple data objects.
Dependencies are inferred from the claim text.