Method and apparatus for efficient memory replication for high availability (HA) protection of a virtual machine (VM)

• US 8,413,145 B2
• Filed: 09/30/2010
• Issued: 04/02/2013
• Est. Priority Date: 09/30/2010
• Status: Active Grant

• Current Assignee: Avaya Incorporated
• Original Assignee: Avaya Incorporated
• Inventors: Zhao, Zhi Qiang, Liu, Feng, Guo, Weiping, Chou, Wu
• Primary Examiner(s): Puente, Emerson
• Assistant Examiner(s): KIM, DONG U
• Application Number:
  US12/895,230
• Publication Number:
  US 20120084782A1
• Time in Patent Office: 915 Days
• Field of Search: None
• US Class Current: 718/1
• CPC Class Codes: G06F 11/1484 involving virtual machines
• CPC Class Codes: G06F 11/2038 with a single idle spare processing component
• CPC Class Codes: G06F 11/2048 where the redundant components share neither address space nor persistent storage
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 12/08 in hierarchically structured memory systems, e.g. virtual memory systems
• CPC Class Codes: G06F 2212/151 Emulated environment, e.g. virtual machine
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors

Abstract

High availability (HA) protection is provided for an executing virtual machine. At a checkpoint in the HA process, the active server suspends the virtual machine; and the active server copies dirty memory pages to a buffer. During the suspension of the virtual machine on the active host server, dirty memory pages are copied to a ring buffer. A copy process copies the dirty pages to a first location in the buffer. At a predetermined benchmark or threshold, a transmission process can begin. The transmission process can read data out of the buffer at a second location to send to the standby host. Both the copy and transmission processes can operate substantially simultaneously on the ring buffer. As such, the ring buffer cannot overflow because the transmission process continues to empty the ring buffer as the copy process continues. This arrangement allows for smaller buffers and prevents buffer overflows.

Claims

1. A method for providing high availability (HA) protection to a virtual machine (VM) executing on an active host, the method comprising:

at a checkpoint interval time, a first processor of an active host executing a copy thread;

the copy thread copying a dirty memory page into a ring buffer;

after the copy thread has copied the dirty memory page into the ring buffer, the first processor of the active host executing a send thread; and

the send thread sending the dirty memory page in the ring buffer to a standby host, wherein the copy thread and the send thread execute on the ring buffer simultaneously;

after initially synchronizing a disk image of the VM by copying the disk image of the VM from a disk storage at the active host to the standby host and while maintaining the synchronization of the disk image of the VM, the first processor synchronizing a memory image of the VM between the active host and the standby host, wherein, after memory synchronization, the memory image of the VM and the disk image of the VM are synchronized;

once disk synchronization and memory synchronization are completed and the disk image and the memory image of the VM are synchronized, a second processor enabling a dynamic disk barrier process to be applied at a later checkpoint time interval; and

at a checkpoint interval time after the later checkpoint time interval, the first processor and the second processor synchronizing the disk image and the memory of the VM between the active host and the standby host for all changes, wherein the standby host controls a synchronization control process at the standby host that utilizes the dynamic disk barrier to keep the memory and the disk image of the VM synchronized at the checkpoint interval for the VM on the active host and the standby host.

2. The method as defined in claim 1, wherein the copy thread copies at least one other dirty memory page into the ring buffer, and wherein the method further comprising:

the first processor determining if a threshold has been reached in the ring buffer;

if the threshold has been reached, the first processor executing the send thread.

3. The method as defined in claim 2, wherein the threshold is a portion of a total capacity of the ring buffer.

4. The method as defined in claim 3, wherein the threshold is 50% of the total capacity.

5. The method as defined in claim 2, wherein the copy thread copies the dirty memory page in a first section of the ring buffer and copies the at least one other dirty memory page into at least one subsequent section of the ring buffer.

6. The method as defined in claim 5, wherein if the threshold has been reached in the ring buffer, the send thread starts sending the dirty memory page in the first section and then the at least one other dirty memory page in the at least one subsequent section of the ring buffer.

7. The method as defined in claim 6, wherein the send thread chases the copy thread.

8. The method as defined in claim 5, wherein determining if the threshold has been reached in the ring buffer is based on storing the dirty memory page in one of the subsequent sections of the ring buffer.

9. The method as defined in claim 1, wherein the send thread is throttled to process data at a speed similar to the copy thread.

10. The method as defined in claim 9, wherein the send thread is throttled to 4 MB per read.

11. A non-transitory computer readable medium having stored thereon instructions that cause a computing system to execute a method for providing high availability (HA) protection to a running virtual machine (VM) executing on an active host, the instructions comprising:

instructions to determine if a checkpoint has arrived during an HA process;

if a checkpoint has arrived, instructions to suspend operation of the VM;

instructions to execute a copy thread that copies two or more dirty memory pages to a ring buffer at the active host;

instructions to determine if a threshold has been reached in the ring buffer, wherein the threshold is associated with data availability in the ring buffer;

instructions to determine if the copy thread has stopped executing; and

if the threshold has been reached or the copy thread has stopped executing, instructions to execute a send thread that sends the two or more dirty memory ages to a standby host, wherein the copy thread and the send thread execute on the ring buffer simultaneously;

after initially synchronizing a disk image of the VM by copying the disk image of the VM from a disk storage at the active host to the standby host and while maintaining the synchronization of the disk image of the VM, instructions to synchronize a memory image of the VM between the active host and the standby host, wherein, after memory synchronization, the memory image of the VM and the disk image of the VM are synchronized;

once disk synchronization and memory synchronization are completed and the disk image and the memory image of the VM are synchronized, instructions to enable a dynamic disk barrier process to be applied at a later checkpoint time interval;

an

at a checkpoint interval time after the later checkpoint time interval, instructions to synchronize the disk image and the memory of the VM between the active host and the standby host for all changes, wherein the standby host controls a synchronization control process at the standby host that utilizes the dynamic disk barrier to keep the memory and the disk image of the VM synchronized at the checkpoint interval for the VM on the active host and the standby host.

12. The computer readable medium as defined in claim 11, further comprising if the copy thread has stopped executing, instructions to resume the VM.

13. The computer readable medium as defined in claim 11, wherein the send thread frees at least one section of the ring buffer for the copy thread to store a dirty memory page.

14. The computer readable medium as defined in claim 13, wherein if the copy thread has stored a dirty memory page in a last section of the ring buffer and still needs to store at least one other dirty memory page, further comprising the copy thread storing a next dirty memory page in a first section of the ring buffer freed by the send thread.

15. A server computing system comprising:

an active host, the active host comprising;

a main memory operable to store a memory image associated with a virtual machine (VM);

a ring buffer operable to store one or more dirty memory pages from the main memory;

a first processor in communication with the main memory and the ring buffer, the first processor executing the VM, the first processor operable to execute a memory replication module, the memory replication module operable to;

execute a copy thread that copies two or more dirty memory pages to the ring buffer;

determine if a threshold has been reached in the ring buffer, wherein the threshold is associated with a portion of a capacity of the ring buffer; and

if the threshold has been reached, instructions to execute a send thread that sends the two or more dirty memory ages to a standby host, wherein;

the send thread chases the copy thread;

the send thread and executes simultaneously as the copy thread; and

the send thread executes on a different section of the ring buffer than the copy thread;

after initially synchronizing a disk image of the VM by copying the disk image of the VM from a disk storage at the active host to the standby host and while maintaining the synchronization of the disk image of the VM, synchronize a memory image of the VM between the active host and the standby host, wherein, after memory synchronization, the memory image of the VM and disk image of the VM are synchronized;

a second processor operable to;

once disk synchronization and memory synchronization are completed and the disk image and the memory image of the VM are synchronized, enable a dynamic disk barrier process to be applied at a checkpoint time interval; and

the first and second processors operable to;

at a checkpoint interval time thereinafter, synchronize the disk image and the memory of the VM between the active host and the standby host for all changes, wherein the standby host controls a synchronization control process at the standby host that utilizes the dynamic disk barrier to keep the memory and disk image of the VM synchronized at the checkpoint interval for the VM on the active host and the standby host.

16. The server computing system as defined in claim 15, further comprising:

the standby host, the standby host comprising;

a main memory, the main memory operable to store a memory image associated with a virtual machine (VM), wherein the main memory receives the dirty memory pages from the send thread.

17. The server computing system as defined in claim 15, wherein after the copy thread has stored dirty memory page data into a last section of the ring buffer, the copy thread is operable to return to a first section of the ring buffer to store a next dirty memory page data.

18. The server computing system as defined in claim 17, wherein the send thread had previously freed the first section of the ring buffer for the copy thread.

19. The server computing system as defined in claim 15, wherein after the copy thread has copied a last dirty memory page to the ring buffer, the VM resumes executing and the send thread continues to send dirty memory page data to a standby host until all dirty memory page data has been sent.