CONTAINER VIRTUAL MACHINES FOR HADOOP

• US 20150120928A1
• Filed: 10/24/2013
• Published: 04/30/2015
• Est. Priority Date: 10/24/2013
• Status: Active Grant

• Current Assignee: VMware, Inc. (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: GUMMARAJU, Jayanth, MCDOUGALL, Richard, NELSON, Michael, GRIFFITH, Rean, DU, Junping, MAGDON-ISMAIL, Tariq, CHEVERESAN, Razvan
• Granted Patent: US 10,193,963 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06F 16/182 Distributed file systems
• CPC Class Codes: G06F 2209/5011 Pool
• CPC Class Codes: G06F 9/5072 Grid computing
• 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 distributed computing application is described that provides a highly elastic and multi-tenant platform for Hadoop applications and other workloads running in a virtualized environment. Data and compute nodes are separated into different virtual machines (VM). Compute VMs are used to launch containers from different tenants. Compute VMs are organized in pools of hot spare VMs that are immediately available for launching a container and executing a task, and pools of cold spare VMs. Each compute VM may include a mounted network filesystem provided by a node manager to share intermediate outputs across VMs executing on the same host.

Claims

1. A method for executing a job comprising a plurality of tasks executed in parallel on a distributed computing application, the method comprising:

generating a list of available virtual machines (VMs) in a powered on state and executing on a first host;

responsive to receiving a request to execute a first task of the plurality of tasks associated with a first tenant on the first host, allocating a first VM from the list of available VMs;

modifying the first VM to mount a network filesystem shared by VMs executing on the first host and associated with the first tenant; and

launching, in the first VM, a container process that executes the first task of the plurality of tasks associated with the first tenant.

2. The method of claim 1, wherein generating the list of available VMs in a powered on state and executing on the first host further comprises:

provisioning a first plurality of VMs to execute on the first host; and

powering on a subset of the first plurality of VMs based on a target pool size.

3. The method of claim 1, wherein allocating the first VM from the list of available VMs further comprises:

allocating the first VM from a first list of available VMs in a powered on state and associated with the first tenant; and

responsive to determining there are no VMs available in the first list of available VMs, allocating the first VM from a second list of available VMs in a powered on state and that have been reset subsequent to a previous task execution.

4. The method of claim 3, further comprising:

responsive to determining there are no VMs available in the second list of available VMs, powering on the first VM from a third list of provisioned VMs in a powered off state.

5. The method of claim 1, further comprising:

generating intermediate output based on execution of the first task, wherein the first task comprises a Map task; and

storing the intermediate output in the mounted network filesystem associated with the first tenant;

executing a web server configured to access contents stored in the mounted network filesystem including the generated intermediate output;

allocating a second VM from the list of available VMs responsive to a second request to execute a second task of the plurality of tasks; and

receiving, by operation of the second VM, the intermediate output results from the web server, wherein the second task comprises a Reduce Task.

6. The method of claim 1, further comprising:

upon completing execution of the first task on the first VM, adding the first VM to the list of available VMs in a powered on state and executing on the first host.

7. The method of claim 6, wherein the list of available VMs in a powered on state and executing on the first host comprises a tenant-specific list of available VMs associated with the first tenant.

8. The method of claim 1, further comprising:

subsequent to execution of the first task on the first VM, performing a reset operation on the first VM that removes tenant-specific state from the first VM; and

adding the first VM to a list of available VMs that have been reset subsequent to executing a task.

9. The method of claim 1, further comprising:

removing the first VM from the list of available VMs; and

responsive to determining the list of available VMs has less than a target pool size, powering on a second VM of a first plurality of VMs provisioned to executed on the first host based on a deficit of the target pool size;

adding the second VM to the list of available VMs.

10. A non-transitory computer-readable storage medium comprising instructions that, when executed in a computing device, execute a job comprising a plurality of tasks executed in parallel on a distributed computing application, by performing the steps of:

generating a list of available virtual machines (VMs) in a powered on state and executing on a first host;

responsive to receiving a request to execute a first task of the plurality of tasks associated with a first tenant on the first host, allocating a first VM from the list of available VMs;

modifying the first VM to mount a network filesystem shared by VMs executing on the first host and associated with the first tenant; and

launching, in the first VM, a container process that executes the first task of the plurality of tasks associated with the first tenant.

11. The non-transitory computer-readable storage medium of claim 10, wherein the step of generating the list of available VMs in a powered on state and executing on the first host further comprises:

provisioning a first plurality of VMs to execute on the first host; and

powering on a subset of the first plurality of VMs based on a target pool size.

12. The non-transitory computer-readable storage medium of claim 10, wherein the step of allocating the first VM from the list of available VMs further comprises:

allocating the first VM from a first list of available VMs in a powered on state and associated with the first tenant; and

responsive to determining there are no VMs available in the first list of available VMs, allocating the first VM from a second list of available VMs in a powered on state and that have been reset subsequent to a previous task execution.

13. The non-transitory computer-readable storage medium of claim 12, further comprising steps of:

responsive to determining there are no VMs available in the second list of available VMs, powering on the first VM from a third list of provisioned VMs in a powered off state.

14. The non-transitory computer-readable storage medium of claim 10, further comprising the steps of:

upon completing execution of the first task on the first VM, adding the first VM to the list of available VMs in a powered on state and executing on the first host.

15. The non-transitory computer-readable storage medium of claim 14, wherein the list of available VMs in a powered on state and executing on the first host comprises a tenant-specific list of available VMs associated with the first tenant.

16. The non-transitory computer-readable storage medium of claim 10, further comprising the steps of:

subsequent to execution of the first task on the first VM, performing a reset operation on the first VM that removes tenant-specific state from the first VM; and

adding the first VM to a list of available VMs that have been reset subsequent to executing a task.

17. The non-transitory computer-readable storage medium of claim 10, further comprising the steps of:

removing the first VM from the list of available VMs; and

responsive to determining the list of available VMs has less than a target pool size, powering on a second VM of a first plurality of VMs provisioned to executed on the first host based on a deficit of the target pool size;

adding the second VM to the list of available VMs.

18. A computer system having a plurality of hosts executing a plurality of virtual machines (VMs) for executing a job comprising a plurality of tasks executed in parallel on a distributed computing application, the computer system comprising:

a memory; and

a processor programmed to carry out the steps of;

generating a list of available virtual machines (VMs) in a powered on state and executing on a first host;

responsive to receiving a request to execute a first task of the plurality of tasks associated with a first tenant on the first host, allocating a first VM from the list of available VMs;

modifying the first VM to mount a network filesystem shared by VMs executing on the first host and associated with the first tenant; and

launching, in the first VM, a container process that executes the first task of the plurality of tasks associated with the first tenant.

19. The computer system of claim 18, wherein the processor configured to allocate the first VM from the list of available VMs is further programmed to carry out the steps of:

allocating the first VM from a first list of available VMs in a powered on state and associated with the first tenant;

responsive to determining there are no VMs available in the first list of available VMs, allocating the first VM from a second list of available VMs in a powered on state and that have been reset subsequent to a previous task execution; and

responsive to determining there are no VMs available in the second list of available VMs, powering on the first VM from a third list of provisioned VMs in a powered off state.

20. The computer system of claim 18, wherein the processor is further configured to carry out the steps of:

upon completing execution of the first task on the first VM, adding the first VM to a tenant-specific list of available VMs that are associated with the first tenant, in a powered on state, and executing on the first host.