Cloud compute scheduling using a heuristic contention model

Patent metadata

Field	Value
Publication number	US-9614779-B2
Application number	US-201314368349-A
Country	US
Kind code	B2
Filing date	Dec 24, 2013
Priority date	Dec 24, 2013
Publication date	Apr 4, 2017
Grant date	Apr 4, 2017

Abstract

Technologies for contention-aware cloud compute scheduling include a number of compute nodes in a cloud computing cluster and a cloud controller. Each compute node collects performance data indicative of cache contention on the compute node, for example, cache misses per thousand instructions. Each compute node determines a contention score as a function of the performance data and stores the contention score in a cloud state database. In response to a request for a new virtual machine, the cloud controller receives contention scores for the compute nodes and selects a compute node based on the contention score. The cloud controller schedules the new virtual machine on the selected compute node. The contention score may include a contention metric and a contention score level indicative of the contention metric. The contention score level may be determined by comparing the contention metric to a number of thresholds. Other embodiments are described and claimed.

First claim

What is claimed is:

1. A cloud controller of a cloud computing cluster, the cloud controller comprising:
a processor;
a compute service module to receive a request for a new virtual machine;
a communication module to receive a plurality of contention scores for a plurality of compute nodes of the cloud computing cluster, wherein each contention score of the plurality of contention scores indicates cache memory contention of a corresponding compute node of the plurality of compute nodes; and
a scheduler module to select a compute node from the plurality of compute nodes based on the corresponding contention score;
wherein the compute service module is further to schedule the new virtual machine on the selected compute node; and
wherein the contention score comprises a contention score level indicative of a contention metric of the compute node, wherein the contention score level comprises a categorization of a magnitude of the contention metric, the contention score level being determined by a comparison of a rate of cache misses of the compute node per reference number of instructions of the compute node to one or more threshold values, and wherein the contention score level comprises: (i) a low contention level indicative of the rate of cache misses per reference number of instructions being less than a low contention threshold value, (ii) a medium contention level indicative of the rate of cache misses per reference number of instructions being greater than or equal to the low contention threshold value and less than a high contention threshold value, or (iii) a high contention level indicative of the rate of cache misses per reference number of instructions being greater than or equal to the high contention threshold value.

2. The cloud controller of claim 1 , wherein the contention score comprises a contention metric, the contention metric being a function of a number of cache misses of the compute node.

3. The cloud controller of claim 2 , wherein the contention metric comprises a rate of cache misses per reference number of instructions of the compute node.

4. The cloud controller of claim 2 , wherein to select the compute node comprises to sort the plurality of compute nodes based on the contention metric to select a compute node having the lowest contention metric.

5. The cloud controller of claim 1 , wherein the low contention threshold value comprises 3 cache misses per thousand instructions, and wherein the high contention threshold value comprises 50 cache misses per thousand instructions.

6. One or more non-transitory, computer-readable storage media comprising a plurality of instructions that in response to being executed cause a cloud controller of a cloud computing cluster to:
receive a request for a new virtual machine;
receive a plurality of contention scores for a plurality of compute nodes of the cloud computing cluster, wherein each contention score of the plurality of contention scores indicates cache memory contention of a corresponding compute node of the plurality of compute nodes;
select a compute node from the plurality of compute nodes based on the corresponding contention score; and
schedule the new virtual machine on the selected compute node;
wherein the contention score comprises a contention score level indicative of a contention metric of the compute node, wherein the contention score level comprises a categorization of a magnitude of the contention metric, the contention score level being determined by comparing a rate of cache misses of the compute node per reference number of instructions of the compute node to one or more threshold values, and wherein the contention score level comprises: (i) a low contention level indicative of the rate of cache misses per reference number of instructions being less than a low contention threshold value, (ii) a medium contention level indicative of the rate of cache misses per reference number of instructions being greater than or equal to the low contention threshold value and less than a high contention threshold value, or (iii) a high contention level indicative of the rate of cache misses per reference number of instructions being greater than or equal to the high contention threshold value.

7. The one or more non-transitory, computer-readable storage media of claim 6 , wherein to receive the contention score comprises to receive a contention metric, the contention metric being a function of a number of cache misses of the compute node.

8. The one or more non-transitory, computer-readable storage media of claim 7 , wherein to receive the contention metric comprises to receive a rate of cache misses per reference number of instructions of the compute node.

Assignees

• Intel Corp

Inventors

• Sesha Subramony
• Patni Archana
• Narayan Ananth S
• Ganguli Mrittika Mrittika

Classifications

• H04L43/16

  Threshold monitoring · CPC title

• H04L67/1097

  for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS] · CPC title

• G06F9/45504

  Abstract machines for programme code execution, e.g. Java virtual machine [JVM], interpreters, emulators · CPC title

• G06F9/5027

  the resource being a machine, e.g. CPUs, Servers, Terminals · CPC title

• H04L67/1095

  Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes · CPC title