High availability data replication

• US 9,489,392 B2
• Filed: 01/24/2014
• Issued: 11/08/2016
• Est. Priority Date: 01/24/2014
• Status: Expired due to Fees

• Current Assignee: CA, Inc. (d/b/a CA Technologies) (Broadcom, Inc.)
• Original Assignee: CA, Inc. (d/b/a CA Technologies) (Broadcom, Inc.)
• Inventors: Wang, Xiaopin, Zhang, Haiyang, Shang, Guoxian, Li, Shaorong
• Primary Examiner(s): Vo, Truong
• Application Number:
  US14/163,940
• Publication Number:
  US 20150213051A1
• Time in Patent Office: 1,019 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/178 Techniques for file synchronisation in file systems

Abstract

A data replication system and method is disclosed in which a master file server uses a first filter in a kernel space to intercept file I/O events, send the file I/O events over a first network, for example a local area network, to a second filter in a kernel space of a first replica file server, and submit the file I/O event to a file system program in user space of the master file server to be processed. The second filter records the file I/O event in a memory of the second file server and sends an acknowledge message to the first filter indicating the file I/O event received by the second filter has been recorded in the replica file server. The first filter notifies user land in the first file server that the file I/O event has been processed if the first filter has received the acknowledge message and the file system program has completed processing the file I/O event in the first file server. In one embodiment, a data replication engine in the first replica file server replicates data stored in the first replica file server to a second replica file server in a location remote from the second replica file server, using a second network, for example a wide area network.

Claims

1. A data replication method, comprising:

intercepting a file I/O in a first kernel space of a first file server before submitting the file I/O to a first file system program operating in a first user space of the first file server;

sending, from the first kernel space, the file I/O as a file I/O event to a second kernel space of a second file server;

submitting, the file I/O to the first file system program without waiting for an acknowledge message from the second file server;

receiving the acknowledge message from the second file server, the acknowledge message indicating the file I/O event has been recorded at the second file server; and

notifying an application or process in the first user space that the file I/O has been successfully processed, upon the condition that the acknowledge message is received by the first file server and the first file system program has completed processing the file I/O.

2. A method according to claim 1 further comprising the second kernel recording the file I/O event received from the first kernel into a memory of the second file server and communicating the acknowledge message back to the first file server.

3. A method according to claim 2, further comprising:

submitting, in a thread program in the second file server kernel space, the file I/O event recorded in the second file server memory to a second file system program at the second file server; and

committing the file I/O event to a storage device associated with the second file server.

4. The data replication method of claim 1, wherein a first filter in the first kernel space of the first file server intercepts the first file I/O.

5. The data replication method of claim 1, wherein sending comprises sending the file I/O event to a socket of the second file server, wherein the socket is associated with a second filter in the second kernel space of the second file server.

6. The data replication method of claim 1, further comprising resending, from the first kernel space of the first file server, the file I/O event to the second kernel space of the second file server after the first file system program has completed processing the file I/O until the receipt of the acknowledge message.

7. A continuous data protection system, comprising:

a first computing system including a first operating system, a first memory, a first data storage device, wherein the first operating system maintains in the first memory a first kernel space to support a first kernel and a first data protection service that runs in the first kernel space, the data protection service to,

send a first file I/O as a first file I/O event to a second computing system,

submit the first file I/O to a first file system program of a first user space without waiting for an acknowledge message from the second computing system, and

indicate to an application or process in the first user space completion of the first I/O dependent upon receipt of an acknowledge message from the second computing system and the first file system program completing processing of the first file I/O;

the second computing system including a second operating system, a second memory, a second data storage device, wherein the second operating system maintains in the second memory a second kernel space to support a second kernel and a second data protection service that runs in the second kernel space, the second data protection service to, listen for file I/O events on a socket,

record the first file I/O event into the second memory after receipt of the first file I/O event, and

communicate an acknowledge message for the first file I/O event to the first computing system after recording the first file I/O event into the second memory.

8. A continuous data protection system according to claim 7, wherein the second data protection service submits the first file I/O event to a data replication engine on the second computing system to replicate to a third computing system.

9. A continuous data protection system according to claim 7, wherein a first network connects the first and second computing systems, the continuous data protection system further comprising a third computing system connected to the second computing system with a second network, the first network having a higher bandwidth than the second network, the second and third computing systems each including a data protection service to replicate data from the second computing system to the third computing system.

10. A continuous data protection system according to claim 7, wherein in case of failure of the second data protection service to send an acknowledge message, the first data protection service to resend the first file I/O as the first file I/O event to the second data protection service.

11. The continuous data protection system of claim 7, wherein the second computing system comprises a second file system program and the second data protection service to submit the first file I/O event to the second file system program after recording the first file I/O event into the second memory.

12. The continuous data protection system of claim 11, further comprising a thread in the second kernel space of the second file server to submit the file I/O event to the second file system program.

13. The continuous data protection system of claim 7, further comprising a first filter in the first kernel space of the first file server to intercept the first file I/O to allow sending the first file I/O before submission to the first file system program.

14. The continuous data protection system of claim 7, further comprising a second filter in the second kernel space of the second file server to receive the file I/O event sent by the first kernel space of the first file server.

15. One or more non-transitory machine-readable media comprising program code to:

in a first kernel space of a first file server, intercept a file I/O event before submitting the file I/O event to a first file system program operating in a first user space of the first file server;

send the file I/O event to a second file server,

submit the file I/O event to the first file system program; and

notify an application or process in the first user space in the first file server that the file I/O event has been successfully processed if an acknowledge message for the file I/O event is received from the second file server and the first file system program has completed processing the file I/O event in the first file server.

16. The one or more non-transitory machine-readable media of claim 15, wherein the program code to intercept comprises program code to intercept with a filter in the first kernel space of the first file server.

17. The one or more non-transitory machine-readable media of claim 15, wherein the program code to send the file I/O event comprises program code to send the file I/O even to a socket in a second kernel space of the second file server.