Customized memory modules in multi-tenant service provider systems

• US 10,491,667 B1
• Filed: 03/26/2015
• Issued: 11/26/2019
• Est. Priority Date: 03/16/2015
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Frink, Darin Lee, Clore, Brent Kenneth
• Primary Examiner(s): Lazaro, David R
• Assistant Examiner(s): Chang, Julian
• Application Number:
  US14/670,246
• Time in Patent Office: 1,706 Days
• CPC Class Codes: G06F 12/023 Free address space management
• CPC Class Codes: G06F 2009/4557 Distribution of virtual machine instances; Migration and load balancing
• CPC Class Codes: G06F 9/45545 Guest-host, i.e. hypervisor is an application program itself, e.g. VirtualBox
• CPC Class Codes: G06F 9/5016 the resource being the memory
• 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: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1021 based on client or server locations
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A computing system providing virtual computing services may maintain a fleet of servers that host virtual machine instances having a wide variety of types and configurations. A service provider may rent processor and memory capacity by defining and offering various virtual machine instances to clients. Each virtual machine instance may include one or more virtual CPUs and a fixed amount of virtualized memory allocated to each virtual CPU, dependent on a predefined ratio between virtual CPU capacity and virtualized memory capacity for the instance type. Each server may include a custom, non-standard sized physical memory module containing memory devices of multiple technologies, types, or sizes on the same printed circuit board. By including custom memory modules, rather than relying only on standard memory modules, the service provider system may implement virtual machines having finer grained options for processor and memory capacity combinations, and may avoid stranding rentable resources.

Claims

1. A system, comprising:

a plurality of computing nodes that implement a hardware virtualization service on behalf of one or more service clients, each of the computing nodes comprising at least one physical processor and a plurality of physical memory modules having a total memory size;

a plurality of virtualized computing resource instances, each hosted on a respective one of the computing nodes and comprising one or more virtual processors and virtualized memory;

wherein the plurality of virtualized computing resource instances comprises one or more instances of each of a plurality of different virtualized computing resource instance types;

wherein the amount of virtualized memory in each virtualized computing resource instance is dependent on a respective ratio between the number of virtual processors and the amount of memory that is defined for its virtualized computing resource instance type; and

wherein, for at least one of the virtualized computing resource instances, the virtualized memory in the at least one virtualized computing resource instance is implemented by one or more of the physical memory modules of the computing node on which it is hosted, wherein the one or more physical memory modules comprise a physical memory module, used in the implementation of the virtualized memory in the at least one virtualized computing resource instance, on which the number of bytes of physical memory does not correspond to a power of two, in order to reduce an amount of stranded physical memory.

2. The system of claim 1,

wherein the hardware virtualization service is configured to support multi-tenancy;

and

wherein the hardware virtualization service provides virtualized computing resource instances for lease by service clients.

3. The system of claim 1,

wherein the hardware virtualization service provides virtualized computing resource instances for lease by service clients;

and

wherein, for one or more of the virtualized computing resource instances, including the at least one of the virtualized computing resource instances, all of the processor capacity of the physical processors that implement the virtual processors and all of the memory capacity of the one or more physical memory modules that are used in the implementation of the virtualized memory are available for lease by service clients within the virtualized computing resource instances at the ratios defined for their types.

4. The system of claim 1, wherein each of the computing nodes hosts virtualized resource instances of a single virtualized computing resource instance type.

5. The system of claim 1, wherein at least one of the computing nodes hosts virtualized resource instances of two or more different virtualized computing resource instance types.

6. A method, comprising:

performing, by one or more computers;

provisioning physical resources of the one or more computers to create a plurality of virtual computing resource instances that are hosted on a computing node comprising the one or more computers for the benefit of one or more clients, wherein the computing node comprises one or more physical memory modules comprising a total memory;

wherein, for each of the plurality of virtual computing resource instances, said provisioning comprises;

provisioning one or more processor cores or hardware execution contexts thereof as virtual processors of the virtual computing resource instance; and

configuring one or more of the one or more physical memory modules to provide virtualized memory for the virtual computing resource instance, wherein the amount of virtualized memory provided for the virtual computing resource instance is dependent on a predetermined ratio between processor capacity and memory capacity; and

wherein, for at least one of the one or more physical memory modules configured to provide virtualized memory for at least one of the plurality of virtual computing resource instances, the number of bytes of physical memory does not correspond to a power of two in order to reduce an amount of stranded physical memory.

7. The method of claim 6,

wherein, for at least two of the plurality of virtual computing resource instances, said provisioning comprises provisioning physical resources to provide different processor capacities or different memory capacities for each of the virtual computing resource instances.

8. The method of claim 6,

wherein the plurality of virtual computing resource instances comprises one or more instances of each of a plurality of different virtual computing resource instance types;

and

wherein the predetermined ratio between processor capacity and memory capacity is different for at least two of the plurality of different virtual computing resource instance types.

9. The method of claim 8,

wherein the method further comprises determining, prior to said provisioning, the plurality of different virtual computing resource instance types for which physical resources are to be provisioned or the number of virtual computing resource instances of each of the plurality of different virtual computing resource instance types for which physical resources are to be provisioned;

and

wherein said determining is dependent on data representing a history of virtual computing resource instance requests or usage, or on a prediction of future virtual computing resource instance demand.

10. The method of claim 6,

wherein the amount of memory on the physical memory module is x gigabytes, where x is a positive integer and not a power of two.

11. The method of claim 6, further comprising:

receiving, from a given client, a request for a virtual computing resource instance, wherein the request specifies a virtual computing resource instance type and a processor capacity; and

assigning to the given client, in response to said receiving, use of a particular one of the plurality of virtual computing resource instances.

12. The method of claim 11, wherein the memory capacity of the physical memory module that provides the virtualized memory for the virtual computing resource instance satisfies a predetermined ratio between processor capacity and memory capacity that is defined for virtual computing resource instances of the specified virtual computing resource instance type having the specified processor capacity.

13. The method of claim 11, wherein the memory capacity of the physical memory module that provides the virtualized memory for the virtual computing resource instance is a physical memory module among available physical memory modules of the plurality of physical resources that most closely matches a memory capacity that satisfies a predetermined ratio between processor capacity and memory capacity that is defined for virtual computing resource instances of the specified virtual computing resource instance type having the specified processor capacity.

14. The method of claim 11, wherein the combined memory capacity of two physical memory modules that provide the virtualized memory for the virtual computing resource instance satisfies a predetermined ratio between processor capacity and memory capacity that is defined for virtual computing resource instances of the specified virtual computing resource instance type having the specified processor capacity.

15. A system, comprising:

one or more computing devices, at least one of which implements a server, wherein the server comprises one or more physical memory modules;

a virtual computing service, implemented by the one or more computing devices;

wherein the server is configured to host a virtual resource instance on behalf of a subscriber or customer of the virtual computing service;

wherein the virtual resource instance comprises;

one or more virtual processors, each implemented by a processor core or hardware execution context of a physical processor; and

virtualized memory implemented by a plurality of integrated circuit memory devices on one or more of the one or more physical memory modules;

wherein the amount of virtualized memory is dependent on a predetermined ratio between virtual processor capacity and virtualized memory capacity; and

wherein the number of bytes of physical memory provided by the integrated circuit memory devices, on at least one of the one or more physical memory modules that implement the virtualized memory of the virtual resource instance, is not a power of two in order to reduce an amount of stranded physical memory.

16. The system of claim 15,

wherein the amount of memory provided by the integrated circuit memory devices on at least one of the one or more physical memory modules is 10 gigabytes (GB), 12 GB, 14 GB, 20 GB, 24 GB, or 36 GB.

17. The system of claim 15,

wherein the physical processor includes multiple memory channels over which it communicates with the one or more physical memory modules;

and

wherein the amount of memory accessed over each of the multiple memory channels is the same.

18. The system of claim 15,

wherein the amount of memory provided by the integrated circuit memory devices on at least one of the one or more physical memory modules is x gigabytes, where x is a positive integer and not a power of two.

19. The system of claim 15,

wherein the amount of memory provided by the integrated circuit memory devices on at least one of the one or more physical memory modules is x terabytes, where x is a positive integer and not a power of two.

20. The system of claim 15,

wherein another one of the computing devices implements another server;

wherein the other server is configured to host another virtual resource instance;

wherein the other virtual resource instance comprises;

one or more virtual processors, each implemented by a processor core or hardware execution context of a physical processor; and

virtualized memory implemented by a plurality of integrated circuit memory devices on one or more physical memory modules; and

wherein the capacity of the one or more virtual processors of the other virtual resource instance or the amount of virtualized memory of the other virtual resource instance is different than that of the virtual resource instance.