System and method for joining containers running on multiple nodes of a cluster

• US 9,621,643 B1
• Filed: 07/31/2015
• Issued: 04/11/2017
• Est. Priority Date: 07/31/2015
• Status: Active Grant

• Current Assignee: Virtuozzo International GmbH
• Original Assignee: Parallels Holdings Limited (HNA Technology Co. Ltd.)
• Inventors: Emelyanov, Pavel
• Primary Examiner(s): Donabed, Ninos
• Application Number:
  US14/814,576
• Time in Patent Office: 620 Days
• Field of Search: None
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors
• CPC Class Codes: G06F 9/5088 involving task migration
• CPC Class Codes: H04L 41/0896 Bandwidth or capacity management, i.e. automatically increasing or decreasing capacities flow or congestion control using dynamic resource allocation, e.g. in-call renegotiation, H04L47/76
• CPC Class Codes: H04L 41/0897 by horizontal or vertical scaling of resources, or by migrating entities, e.g. virtual resources or entities
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30

Abstract

A method for joining Containers and balancing loads uses a cluster of independent computer nodes assembled into a cloud service. The cluster nodes have Containers implemented on them. The Containers execute some real processes that use the resources of the node. If the Container requires additional resources (CPU, memory, etc.), the Container is provided with resources available from another node. In order to do this, the process requiring additional resources is migrated to another Container on a different node, and a proxy process imitates this process execution on the original Container. A load balancer monitors the resource usage and selects the Containers to be joined.

Claims

1. A computer-implemented method for joining containers running on a cluster of hardware nodes, the method comprising:

on a plurality of nodes connected into a cluster, starting a plurality of containers,

where each node hosts at least one container;

executing a plurality of real processes in the containers, where each real process is executed on a single container and other containers execute fake program modules corresponding to the real process;

determining the real process on a source container to be migrated and a target container where the real process is to be migrated;

for the target container, replacing the fake program module with the real process and replacing the real process with the fake program module on the source container; and

re-launching a process on the same hardware node with live settings from another node,

wherein the fake program module includes interfaces identical to interfaces of a corresponding real process;

wherein interfaces of real processes are connected to corresponding interfaces of the corresponding fake program modules so as to establish output data flow of the corresponding interfaces of corresponding fake program modules identical to the data flow outputted by interfaces of a corresponding real process; and

wherein interfaces of the fake program modules are configured to receive data from conjugated interfaces.

2. The method of claim 1, further comprising disconnecting interfaces of the real processes from corresponding interfaces of the fake processes when the interfaces of real processes are connected to interfaces of other real processes of the same container.

3. The method of claim 1, further comprising discarding data flows directed from the real processes to the fake program modules received at the interfaces of the fake program modules connected to interfaces of another program module.

4. The method of claim 1, further comprising storing data of the real processes.

5. The method of claim 1, further comprising applying a check point method to distribution of the real processes and the fake program modules for minimizing computational losses.

6. The method of claim 5, wherein the check point method is Checkpoint/currently amended in User space (CRIU) method.

7. The method of claim 5, wherein the checkpoint method provides for the re-launching of the process.

8. A system for joining containers running on a cluster of hardware nodes, the system comprising:

a plurality of nodes connected into a cluster, each node having a processor;

a plurality of containers running on the nodes, wherein each node hosts at least one container;

a plurality of real processes executing in the containers, wherein each real process is executed on a single container and other containers execute fake program modules corresponding to the real process on their respective processors;

wherein the system determines the real process on a source container to be migrated and a target container where the real process is to be migrated;

for the target container, the system replaces the fake program module with the real process and replaces the real process with the fake program module on the source container; and

the system re-launches a process on the same hardware node with live settings from another node,

wherein the fake program module includes interfaces identical to interfaces of a corresponding real process;

wherein interfaces of real processes are connected to a corresponding interfaces of the corresponding fake program modules so as to establish output data flow of the corresponding interfaces of corresponding fake program modules identical to the data flow outputted by interfaces of a corresponding real process; and

wherein interfaces of the fake program modules receive data from conjugated interfaces.

9. The system of claim 8, wherein interfaces of the real processes are disconnected from corresponding interfaces of the fake processes when the interfaces of real processes are connected to interfaces of other real processes of the same container.

10. The system of claim 8, wherein data flows directed from the real processes to the fake program modules are discarded when the data flows are received at the interfaces of the fake program modules connected to interfaces of another program module.

11. The system of claim 8, wherein the system applies a check point method to distribution of the real processes and the fake program modules for minimizing computational losses.

12. The system of claim 11, wherein the check point method is Checkpoint/Restore in User space (CRIU) method.

13. The system of claim 11, wherein the checkpoint method provides for the re-launching of the process.

14. A non-transitory computer storage medium storing computer code for implementing a method for joining containers running on a cluster of hardware nodes, the method comprising:

on a plurality of nodes connected into a cluster, starting a plurality of containers,

where each node hosts at least one container;

executing a plurality of real processes in the containers, where each real process is executed on a single container and other containers execute fake program modules corresponding to the real process;

determining the real process on a source container to be migrated and a target container where the real process is to be migrated;

for the target container, replacing the fake program module with the real process and replacing the real process with the fake program module on the source container; and

re-launching a process on the same hardware node with live settings from another node,

wherein the fake program module includes interfaces identical to interfaces of a corresponding real process;

wherein interfaces of real processes are connected to corresponding interfaces of the corresponding fake program modules so as to establish output data flow of the corresponding interfaces of corresponding fake program modules identical to the data flow outputted by interfaces of a corresponding real process; and

wherein interfaces of the fake program modules are configured to receive data from conjugated interfaces.