Cross-platform scheduling with long-term fairness and platform-specific optimization

• US 9,886,307 B2
• Filed: 07/01/2015
• Issued: 02/06/2018
• Est. Priority Date: 11/21/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Hildrum, Kirsten W., Nabi, Zubair, Nagarajan, Viswanath, Saccone, Robert, Sarpatwar, Kanthi K., Wagle, Rohit, Wolf, Joel Leonard
• Primary Examiner(s): Ghaffari, Abu
• Application Number:
  US14/789,024
• Publication Number:
  US 20160147566A1
• Time in Patent Office: 951 Days
• Field of Search: None
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/5011 the resources being hardware resources other than CPUs, Servers and Terminals
• CPC Class Codes: G06F 9/5088 involving task migration
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/104 Peer-to-peer [P2P] networks
• CPC Class Codes: H04L 67/61 taking into account QoS or priority requirements

Abstract

Methods, systems, and computer program products for cross-platform scheduling with fairness and platform-specific optimization are provided herein. A method includes determining dimensions of a set of containers in which multiple tasks associated with a request are to be executed; assigning each of the containers to a processing node on one of multiple platforms based on the dimensions of the given container, and to a platform owner selected from the multiple platforms based on a comparison of resource requirements of each of the multiple platforms and the dimensions of the given container; and generating container assignments across the set of containers by incorporating the assigned node of each container in the set of containers, the assigned platform owner of each container in the set of containers, one or more scheduling requirements of each of the platforms, one or more utilization objectives, and enforcing a sharing guarantee of each of the platforms.

Claims

1. A method comprising the following steps:

determining, within a datacenter, one or more dimensions of a set of containers in which multiple tasks associated with a request are to be executed, wherein the datacenter comprises;

multiple processing nodes and multiple platforms,

wherein the multiple platforms comprise multiple software entities which compete for one or more resources of the processing nodes in the datacenter, and

wherein the set of containers comprises multiple software containers within the processing nodes of the datacenter, wherein each of the containers comprises an allocation of a given fraction of each of the one or more resources of the processing node to which each of the containers is assigned;

assigning each container in the set of containers (i) to one of the processing nodes on one of the multiple platforms based on the determined one or more dimensions of the container being assigned, and (ii) to a platform owner selected from owners of the multiple platforms based on a comparison of one or more resource requirements attributed to each of the multiple platforms and the determined one or more dimensions of the container; and

generating container assignments across the set of containers for the multiple tasks associated with the request, wherein said generating comprises incorporating (i) the assigned node of each container in the set of containers, (ii) the assigned platform owner of each container in the set of containers, (iii) one or more scheduling requirements of each of the multiple platforms, wherein said one or more scheduling requirements comprises one or more parameters for implementing preemption of a container from a given sharing platform, and wherein the one or more parameters for implementing preemption of a container comprises ownership identification of the container, and (iv) one or more utilization objectives comprising at least long-term fairness across the multiple platforms, wherein the long-term fairness (a) increases use of resources across the multiple platforms, and (b) facilitates sharing of resources across the multiple platforms to a predetermined extent embodied via a sharing guarantee of each of the multiple platforms, wherein said sharing guarantee of each of the multiple platforms comprises a self-appointed status representing each given platform'"'"'s degree of willingness to share containers with one or more additional platforms, wherein the degree includes a self-appointed quantity of one or more containers;

wherein said steps are carried out by at least one computing device.

2. The method of claim 1, wherein said determining step is carried out offline.

3. The method of claim 1, wherein said assigning step is carried out offline.

4. The method of claim 1, wherein said generating step is carried out online.

5. The method of claim 1, wherein the one or more dimensions comprise central processing unit cores.

6. The method of claim 1, wherein the one or more dimensions comprises memory.

7. The method of claim 1, wherein the one or more dimensions comprises a fixed number of multiple resource container sizes into which the multiple tasks will fit.

8. The method of claim 1, wherein said determining comprises utilizing, as input, multiple historical work requests and multiple expected work requests associated with the multiple platforms.

9. The method of claim 1, comprising:

maintaining, for each of the multiple platforms, an identification of each of the one or more containers owned by the given platform.

10. The method of claim 1, wherein said generating further comprises incorporating a distinct request priority for each respective one of the multiple tasks associated with the request.

11. The method of claim 1, wherein said generating further comprises incorporating a distinct container preemption priority for each respective one of the multiple tasks associated with the request.

12. The method of claim 1, wherein said one or more scheduling requirements comprise a rack-level constraint and/or a host-level locality constraint.

13. The method of claim 1, wherein said one or more scheduling requirements comprises a type of container to be used to satisfy the request.

14. The method of claim 13, wherein the type of container to be used to satisfy the request comprises only containers owned by a given platform.

15. The method of claim 13, wherein the type of container to be used to satisfy the request comprises unused containers of one or more sharing platforms only after containers owned by a given platform have been exhausted.

16. The method of claim 13, wherein the type of container to be used to satisfy the request comprises only containers not owned by a given platform.

17. The method of claim 1, wherein said implementing preemption comprises (i) selecting the container from the given sharing platform, (ii) notifying the given sharing platform that the given sharing platform has a configured amount of time to release the selected container, and (iii) reassigning allocation of the selected container after the configured amount of time has elapsed.