Disaggregated Fabric-Switched Computing Units

• US 20180046514A1
• Filed: 08/11/2017
• Published: 02/15/2018
• 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
• Granted Patent: US 10,642,659 B2
• Time in Patent Office: 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 compute units is presented that includes forming compute units among a plurality of physical computing components comprising at least central processing units (CPUs), storage modules, and network interface modules coupled over a Peripheral Component Interconnect Express (PCIe) fabric configured to communicatively couple the plurality of physical computing components and isolate the compute unit in the clustered environment using logical partitioning within the PCIe fabric. The method also includes initiating a software component deployed to at least associated CPUs within the compute units, reporting telemetry to the management processor related to operation of the compute unit, and emulating operation of an Ethernet interface to an operating system of the associated CPU for transfer of communications comprising at least the telemetry to the management processor over the PCIe fabric.

Claims

1. A compute unit in a clustered environment, comprising:

a plurality of physical computing components comprising at least one central processing unit (CPU), storage module, and network interface module coupled over a Peripheral Component Interconnect Express (PCIe) fabric that is configured to communicatively couple the plurality of physical computing components and form the compute unit using logical partitioning within the PCIe fabric; and

a software component deployed by a management processor to at least an associated CPU within the compute unit responsive to formation of the compute unit, the software component comprising;

a monitoring function configured to report telemetry to the management processor related to operation of the compute unit; and

a driver function configured to emulate operation of an Ethernet interface to an operating system of the associated CPU for transfer of communications comprising at least the telemetry to the management processor over the PCIe fabric.

2. The compute unit of claim 1, wherein the driver function is further configured to transfer the communications over the PCIe fabric for delivery to the management processor without processing the communications through a network stack.

3. The compute unit of claim 1, wherein based at least on the logical partitioning, the compute unit has visibility over the PCIe fabric to only the plurality of physical computing components assigned to the compute unit using the logical partitioning within the PCIe fabric, and lacks visibility over the PCIe fabric to other physical computing components that are communicatively coupled over the PCIe fabric.

4. The compute unit of claim 1, wherein the management processor is configured to instruct the PCIe fabric to establish the logical partitioning within the PCIe fabric by at least forming domain-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

5. The compute unit of claim 1, wherein the management processor is configured to instruct the PCIe fabric to establish the logical partitioning within the PCIe fabric by at least forming non-transparent (NT) port-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

6. The compute unit of claim 1, comprising:

responsive to alteration of the logical partitioning by the management processor, the compute unit configured to alter composition of the plurality of physical computing components within the compute unit.

7. The compute unit of claim 6, wherein the composition of the plurality of physical computing components within the compute unit is altered to include at least one more central processing unit (CPU), storage module, and network interface module.

8. The compute unit of claim 1, wherein the plurality of physical computing components further comprise a graphics processing unit (GPU).

9. The compute unit of claim 1, wherein the compute unit is a member of a cluster comprising at least one further compute unit coupled over the PCIe fabric and logically isolated on the PCIe fabric from other clusters.

10. A method of operating a compute unit in a clustered environment, the method comprising:

forming the compute unit among a plurality of physical computing components comprising at least a central processing unit (CPU), storage module, and network interface module coupled over a Peripheral Component Interconnect Express (PCIe) fabric that is configured to communicatively couple the plurality of physical computing components and isolate the compute unit in the clustered environment using logical partitioning within the PCIe fabric;

initiating a software component deployed by a management processor to at least an associated CPU within the compute unit responsive to formation of the compute unit;

in a monitoring function of the software component, reporting telemetry to the management processor related to operation of the compute unit; and

in a driver function of the software component, emulating operation of an Ethernet interface to an operating system of the associated CPU for transfer of communications comprising at least the telemetry to the management processor over the PCIe fabric.

11. The method of claim 10, further comprising:

in the driver function, transferring the communications over the PCIe fabric for delivery to the management processor without processing the communications through a network stack.

12. The method of claim 10, wherein based at least on the logical partitioning, the compute unit has visibility over the PCIe fabric to only the plurality of physical computing components assigned to the compute unit using the logical partitioning within the PCIe fabric, and lacks visibility over the PCIe fabric to other physical computing components that are communicatively coupled over the PCIe fabric.

13. The method of claim 10, wherein the management processor is configured to instruct the PCIe fabric to establish the logical partitioning within the PCIe fabric by at least forming domain-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

14. The method of claim 10, wherein the management processor is configured to instruct the PCIe fabric to establish the logical partitioning within the PCIe fabric by at least forming non-transparent (NT) port-based PCIe segregation among ports of PCIe switches that comprise the PCIe fabric.

15. The method of claim 10, further comprising:

responsive to alteration of the logical partitioning by the management processor, changing a composition of the plurality of physical computing components within the compute unit.

16. The method of claim 15, wherein the composition of the plurality of physical computing components within the compute unit is altered to include at least one more central processing unit (CPU), storage module, and network interface module.

17. The method of claim 15, wherein the composition of the plurality of physical computing components within the compute unit is altered to include at least one graphics processing unit (GPU).

18. The method of claim 10, wherein the compute unit is a member of a cluster comprising at least one further compute unit coupled over the PCIe fabric and logically isolated on the PCIe fabric from other clusters of the clustered environment.

19. A computing apparatus comprising:

one or more computer readable storage media;

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

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

form a compute unit among a plurality of physical computing components comprising at least a central processing unit (CPU), storage module, and network interface module coupled over a Peripheral Component Interconnect Express (PCIe) fabric that is configured to communicatively couple the plurality of physical computing components and isolate the compute unit in the clustered environment using logical partitioning within the PCIe fabric;

initiate a software component deployed by a management processor to at least an associated CPU within the compute unit responsive to formation of the compute unit;

in a monitoring function of the software component, report telemetry to the management processor related to operation of the compute unit; and

in a driver function of the software component, emulate operation of an Ethernet interface to an operating system of the associated CPU for transfer of communications comprising at least the telemetry to the management processor over the PCIe fabric without processing the communications through a network stack.

20. The computing apparatus of claim 19, wherein based at least on the logical partitioning, the compute unit has visibility over the PCIe fabric to only the plurality of physical computing components assigned to the compute unit using the logical partitioning within the PCIe fabric, and lacks visibility over the PCIe fabric to other physical computing components that are communicatively coupled over the PCIe fabric.