Hyperscale artificial intelligence and machine learning infrastructure
A hyperscale artificial intelligence and machine learning infrastructure includes a plurality of racks, where: at least one or more of the racks include one or more GPU servers; at least one or more of the racks include one or more storage systems; each of the racks include one or more switches coupled to at least one switch in another rack; and the one or more GPU servers are configured to execute one or more artificial intelligence or machine learning applications, wherein data stored within the one or more storage systems is used as input to the one or more artificial intelligence or machine learning applications.
1. A hyperscale artificial intelligence and machine learning infrastructure, the hyperscale artificial intelligence and machine learning infrastructure including:
a plurality of racks, wherein:
one or more of the racks include one or more GPU (‘Graphical Processor Unit’) servers;
one or more of the racks includes at least one storage system that separates network and data planes to expose one or more storage resources as a single namespace;
one or more of the racks include one or more switches coupled to at least one switch in another rack; and
the one or more GPU servers are configured to:
execute the one or more artificial intelligence or machine learning applications, using input data stored within the one or more storage systems that is accessed using the single namespace.
2. The infrastructure of claim 1 , wherein each of the storage systems comprises a fabric module configured to provision a software defined network for storage resources.
3. The infrastructure of claim 2 , wherein the fabric module of each of the storage systems presents a single namespace for the plurality of racks.
4. The infrastructure of claim 3 , wherein:
the plurality of racks is configured to maintain the single namespace upon coupling of an additional rack for scale out.
5. The infrastructure of claim 1 , wherein the switches and racks are configured in a leaf-spine network topology.
6. The infrastructure of claim 1 , wherein the switches and racks are configured in a torus network topology.
7. The infrastructure of claim 1 , wherein the switches and racks are configured in a hierarchical network topology.
8. The infrastructure of claim 1 , wherein at least one of the plurality of racks comprises:
one or more GPU servers, one or more storage systems, and one or more switches.
9. The infrastructure of claim 1 , wherein:
at least one of the plurality of racks comprises one or more GPU servers, one or more switches, and no storage systems; and
at least one other of the plurality of racks comprises one or more storage systems, one or more switches, and no GPU servers.
10. A method of performing artificial intelligence and machine learning processes in a hyperscale artificial intelligence and machine learning infrastructure, the infrastructure comprising:
a plurality of racks, wherein:
one or more of the racks include one or more GPU (‘Graphical Processor Unit’) servers;
one or more of the racks includes at least one storage system that separates network and data planes to expose one or more storage resources as a single namespace;
one or more of the racks include one or more switches coupled to at least one switch in another rack; and
wherein, the method further comprises executing, by the one or more GPU servers, one or more artificial intelligence or machine learning applications, including utilizing as input to the one or more artificial intelligence or machine learning applications, data stored within the one or more storage systems that is accessed using the single namespace.
11. The method of claim 10 , further comprising,
provisioning, by a fabric module of each of the storage systems, a software defined network for storage resources.
12. The method of claim 11 , further comprising presenting a single namespace for the plurality of racks.
13. The method of claim 12 , further comprising maintaining the single namespace upon coupling of an additional rack for scale out.
14. The method of claim 10 , wherein the switches and racks are configured in a leaf-spine network topology.
15. The method of claim 10 , wherein the switches and racks are configured in a torus network topology.
16. The method of claim 10 , wherein the switches and racks are configured in a hierarchical network topology.
17. The method of claim 10 , wherein at least one of the plurality of racks comprises:
one or more GPU servers, one or more storage systems, and one or more switches.
18. The method of claim 10 , wherein:
at least one of the plurality of racks comprises one or more GPU servers, one or more switches, and no storage systems; and
at least one other of the plurality of racks comprises one or more storage systems, one or more switches, and no GPU servers.