Managing data center resources to achieve a quality of service

Patent metadata

Field	Value
Publication number	US-11722382-B2
Application number	US-202117494569-A
Country	US
Kind code	B2
Filing date	Oct 5, 2021
Priority date	Sep 28, 2012
Publication date	Aug 8, 2023
Grant date	Aug 8, 2023

Abstract

In accordance with some embodiments, a cloud service provider may operate a data center in a way that dynamically reallocates resources across nodes within the data center based on both utilization and service level agreements. In other words, the allocation of resources may be adjusted dynamically based on current conditions. The current conditions in the data center may be a function of the nature of all the current workloads. Instead of simply managing the workloads in a way to increase overall execution efficiency, the data center instead may manage the workload to achieve quality of service requirements for particular workloads according to service level agreements.

First claim

The invention claimed is:

1. A method implemented using a computer system, the computer system being for use in association with dynamic cloud-based workload and/or data center resource allocation, the computer system comprising multiple hardware central processing unit (CPU) cores associated with multiple server nodes, the multiple hardware CPU cores being associated with CPU performance monitoring hardware to generate cache utilization-related data and memory bandwidth utilization-related data associated with execution of workloads by the multiple CPU cores, the method comprising:
collecting the cache utilization-related data and/or the memory bandwidth utilization-related data;
generating, on a socket-by-socket-basis, socket-level workload performance-related data based upon the cache utilization-related data and/or the memory bandwidth utilization-related data; and
dynamically modifying, based upon (1) quality of service data and (2) the cache utilization-related data and/or the memory bandwidth-related data, allocation of the workloads to resources of the multiple CPU cores so as to address socket-level workload resource contention;
wherein:
the dynamically modifying comprises changing at least one workload-to-CPU core placement, based upon the socket-level workload performance-related data, so as to address the socket-level workload resource contention.

2. The method of claim 1 , wherein:
the dynamically modifying also comprises changing at least one workload-to-server-node placement.

3. The method of claim 2 , wherein:
the collecting comprises reading, at regular intervals, CPU hardware performance counters (PMU); and
the allocation of the workloads to the resources is implemented, at least in part, using a cloud-based scheduler.

4. The method of claim 3 , wherein:
the resources of the multiple CPU cores comprise memory bandwidth and/or cache resources; and
the dynamically modifying also comprises changing at least one memory bandwidth allocation and/or at least one cache resource allocation to at least one other of the workloads.

5. The method of claim 4 , wherein:
the quality of service data is based, at least in part, upon service level agreement data; and
one or more of the workloads are associated with one or more virtual machines and/or containers.

6. The method of claim 4 , further comprising:
using memory bandwidth-based hints in allocating of the resources.

7. Computer system for use in association with dynamic cloud-based workload and/or data center resource allocation, the computer system comprising:
multiple hardware central processing unit (CPU) cores associated with multiple server nodes, the multiple hardware CPU cores being associated with CPU performance monitoring hardware to generate cache utilization-related data and memory bandwidth utilization-related data associated with execution of workloads by the multiple CPU cores; and
memory storing instructions to be executed by the computer system, the instructions when executed resulting in the computer system being configured for performance of operations comprising:
collecting the cache utilization-related data and/or the memory bandwidth utilization-related data;
generating, on a socket-by-socket-basis, socket-level workload performance-related data based upon the cache utilization-related data and/or the memory bandwidth utilization-related data; and
dynamically modifying, based upon (1) quality of service data and (2) the cache utilization-related data and/or the memory bandwidth-related data, allocation of the workloads to resources of the multiple CPU cores so as to address socket-level workload resource contention;

wherein:
the dynamically modifying comprises changing at least one workload-to-CPU core placement, based upon the socket-level workload performance-related data, so as to address the socket-level workload resource contention.

8. The computer system of claim 7 , wherein:
the dynamically modifying also comprises changing at least one workload-to-server-node placement.

9. The computer system of claim 8 , wherein:
the collecting comprises reading, at regular intervals, CPU hardware performance counters (PMU); and
the allocation of the workloads to the resources is implemented, at least in part, using a cloud-based scheduler.

10. The computer system of claim 9 , wherein:
the resources of the multiple CPU cores comprise memory bandwidth and/or cache resources; and
the dynamically modifying also comprises changing at least one memory bandwidth allocation and/or at least one cache resource allocation to at least one other of the workloads.

11. The computer system of claim 10 , wherein:
the quality of service data is based, at least in part, upon service level agreement data; and
one or more of the workloads are associated with one or more virtual machines and/or containers.

12. The computer system of claim 10 , wherein:
the operations further comprise using memory bandwidth-based hints in allocating of the resources.

13. Computer system for use in association with dynamic cloud-based workload and/or data center resource allocation, the computer system comprising:
multiple server nodes comprising multiple hardware central processing unit (CPU) cores, the multiple hardware CPU cores being associated with CPU performance monitoring hardware to generate cache utilization-related data and memory bandwidth utilization-related data associated with execution of workloads by the multiple CPU cores;
at least one other node associated with at least one cloud-based scheduler; and
memory storing instructions to be executed by at least certain of the nodes of the computer system, the instructions when executed resulting in the computer system being configured for performance of operations comprising:
collecting the cache utilization-related data and/or the memory bandwidth utilization-related data from the multiple server nodes;
generating, on a socket-by-socket-basis, at least in part by the at least one other node, socket-level workload performance-related data based upon the cache utilization-related data and/or the memory bandwidth utilization-related data; and
dynamically modifying, by the at least one other node, based upon (1) quality of service data and (2) the cache utilization-related data and/or the memory bandwidth-related data, allocation of the workloads to resources of the multiple CPU cores so as to address socket-level workload resource contention;

wherein:
the dynamically modifying comprises changing at least one workload-to-CPU core placement, based upon the socket-level workload performance-related data, so as to address the socket-level workload resource contention.

14. The computer system of claim 13 , wherein:
the dynamically modifying also comprises changing at least one workload-to-server-node placement.

15. The computer system of claim 14 , wherein:
the collecting comprises reading, at regular intervals, CPU hardware performance counters (PMU); and
the allocation of the workloads to the resources is implemented, at least in part, using the cloud-based scheduler.

16. The computer system of claim 15 , wherein:
the resources of the multiple CPU cores comprise memory bandwidth and/or cache resources; and
the dynamically modifying also comprises changing at least one memory bandwidth allocation and/or at least one cache resource allocation to at least one other of the workloads.

17. The computer system of claim 16 , wherein:
the quality of service data is based, at least in part, upon service level agreement data; a

Assignees

• Intel Corp

Inventors

• Ganguli Mrittika
• I Muthuvel M
• Narayan Ananth S
• Moses Jaideep
• Herdrich Andrew J
• Khanna Rahul

Classifications

• H04L41/50 Primary

  Network service management, e.g. ensuring proper service fulfilment according to agreements · CPC title

• H04L41/145

  involving simulating, designing, planning or modelling of a network · CPC title

• H04L41/5009

  Determining service level performance parameters or violations of service level contracts, e.g. violations of agreed response time or mean time between failures [MTBF] · CPC title

• H04L41/5025 Primary

  by proactively reacting to service quality change, e.g. by reconfiguration after service quality degradation or upgrade · CPC title