Disaggregated fabric-switched computing platform

• US 10,592,291 B2
• Filed: 08/11/2017
• Issued: 03/17/2020
• Est. Priority Date: 08/12/2016
• Status: Active Grant

• Current Assignee: Liqid Inc
• Original Assignee: Liqid Inc
• Inventors: Breakstone, Jason, Long, Christopher R., Cannata, James Scott
• Primary Examiner(s): Shyu, Jing-Yih
• Application Number:
  US15/675,377
• Publication Number:
  US 20180046513A1
• Time in Patent Office: 949 Days
• CPC Class Codes: G06F 12/02 Addressing or allocation; Relocation program address sequencing G06F9/00; arrangements for selecting an address in a digital store G11C8/00
• CPC Class Codes: G06F 13/4022 using switching circuits, e.g. switching matrix, connection or expansion network G06F13/4009 takes precedence
• CPC Class Codes: G06F 2203/04803 Split screen, i.e. subdividing the display area or the window area into separate subareas
• CPC Class Codes: G06F 2213/0026 PCI express
• CPC Class Codes: G06F 3/0482 Interaction with lists of selectable items, e.g. menus
• CPC Class Codes: G06F 3/0484 for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
• CPC Class Codes: G06F 9/451 Execution arrangements for user interfaces
• CPC Class Codes: G06F 9/5016 the resource being the memory
• CPC Class Codes: G06F 9/5044 considering hardware capabilities
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources
• CPC Class Codes: G06F 9/5077 Logical partitioning of resources; Management or configuration of virtualized resources specific details on emulation or internal functioning of virtual machines G06F9/455
• CPC Class Codes: G06F 9/54 Interprogram communication

Abstract

Disaggregated computing architectures, platforms, and systems are provided herein. In one example, a method of operating a disaggregated computing architecture is presented. The method includes, receiving user commands to establish compute units among a plurality of physical computing components, each of the compute units comprising one or more of the plurality of physical computing components selected from among central processing units (CPUs), graphics processing units (GPUs), storage modules, and network interface modules. The method also includes forming the compute units based at least on logical partitioning within a Peripheral Component Interconnect Express (PCIe) fabric communicatively coupling the plurality of physical computing components, wherein each of the compute units have visibility over the PCIe fabric to the one or more of the plurality of physical computing components assigned to the associated compute units using the logical partitioning within the PCIe fabric.

Claims

1. A disaggregated computing platform, comprising:

a management processor configured to receive user commands to establish compute units among a plurality of physical computing components, each of the compute units comprising one or more associated physical computing components of the plurality of physical computing components selected from among central processing units (CPUs), graphics processing units (GPUs), storage modules, and network interface modules; and

the management processor configured to instruct a Peripheral Component Interconnect Express (PCIe) fabric communicatively coupling the plurality of physical computing components to establish the logical partitioning within the PCIe fabric and form at least a portion of the compute units using domain-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

2. The disaggregated computing platform of claim 1, comprising:

the management processor configured to instruct the PCIe fabric to form at least another portion of the compute units using non-transparent (NT) port-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

3. The disaggregated computing platform of claim 1, comprising:

the management processor configured to alter composition of the associated physical computing components of the plurality of physical computing components within the compute units by at least changing the logical partitioning within the PCIe fabric responsive to the user commands.

4. The disaggregated computing platform of claim 1, wherein a first compute unit comprises at least one CPU, storage module, and network interface module.

5. The disaggregated computing platform of claim 1, wherein a first compute unit comprises at least one CPU, GPU, storage module, and network interface module.

6. The disaggregated computing platform of claim 1, comprising:

responsive to forming each of the compute units, the management processor configured to deploy a software component to at least an associated CPU within each of the compute units.

7. The disaggregated computing platform of claim 6, wherein the software component comprises a monitoring function configured to report telemetry data to the management processor related to operation of an associated compute unit.

8. The disaggregated computing platform of claim 6, wherein the software component comprises a driver function configured to emulate operation of an Ethernet interface to an operating system of the associated CPU for communication with the management processor over the PCIe fabric.

9. The disaggregated computing platform of claim 8, wherein the driver function is further configured to transfer communications over the PCIe fabric for delivery to the management processor without processing the communications through a network stack.

10. The disaggregated computing platform of claim 1, wherein each of the compute units have visibility over the PCIe fabric to only the one or more associated physical computing components of the plurality of physical computing components assigned to the associated compute units using the logical partitioning within the PCIe fabric.

11. A method of operating a disaggregated computing platform, the method comprising:

in a management processor, receiving user commands to establish compute units among a plurality of physical computing components, each of the compute units comprising one or more associated physical computing components of the plurality of physical computing components selected from among central processing units (CPUs), graphics processing units (GPUs), storage modules, and network interface modules;

in the management processor, instructing a Peripheral Component Interconnect Express (PCIe) fabric communicatively coupling the plurality of physical computing components to establish logical partitioning within the PCIe fabric and form at least a portion of the compute units using domain-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric;

wherein each of the compute units have visibility over the PCIe fabric to the one or more associated physical computing components of the plurality of physical computing components assigned to the associated compute units using the logical partitioning within the PCIe fabric.

12. The method of claim 11, further comprising:

in the management processor, instructing the PCIe fabric to form at least another portion of the compute units using non-transparent (NT) port-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

13. The method of claim 11, further comprising:

altering composition of the associated physical computing components of the plurality of physical computing components within the compute units by at least changing the logical partitioning within the PCIe fabric responsive to the user commands.

14. The method of claim 11, further comprising:

responsive to forming each of the compute units, deploying a software component to at least an associated CPU within each of the compute units.

15. The method of claim 14, wherein the software component comprises a monitoring function configured to report telemetry data to the management processor related to operation of an associated compute unit.

16. The method of claim 14, wherein the software component comprises a driver function configured to emulate operation of an Ethernet interface to an operating system of the associated CPU for communication with the management processor over the PCIe fabric, wherein the driver function is further configured to transfer communications over the PCIe fabric for delivery to the management processor without processing the communications through a network stack.

17. A computing apparatus comprising:

one or more non-transitory computer readable storage media;

a processing system operatively coupled with the one or more non-transitory computer readable storage media; and

program instructions stored on the one or more non-transitory computer readable storage media, that when executed by the processing system, direct the processing system to at least;

receive user commands to establish compute units among a plurality of physical computing components, each of the compute units comprising one or more associated physical computing components of the plurality of physical computing components selected from among a central processing units (CPUs), graphics processing units (GPUs), storage modules, and network interface modules; and

form the compute units based at least on logical partitioning within a Peripheral Component Interconnect Express (PCIe) fabric communicatively coupling the plurality of physical computing components using at least domain-based segregation among ports of switches that comprise, wherein each of the compute units have visibility over the PCIe fabric to the one or more associated physical computing components of the plurality of physical computing components assigned to the associated compute units using the logical partitioning within the PCIe fabric.

18. The computing apparatus of claim 17, comprising further program instructions, when executed by the processing system, direct the processing system to at least:

responsive to forming each of the compute units, deploy a software component to at least an associated CPU within each of the compute units;

wherein the software component comprises a driver function configured to emulate operation of an Ethernet interface to an operating system of the associated CPU for communication with a management processor over the PCIe fabric; and

wherein the driver function is further configured to transfer communications over the PCIe fabric for delivery to the management processor without processing the communications through a network stack.