Seamless takeover of a stateful protocol session in a virtual machine environment

• US 9,600,315 B2
• Filed: 10/22/2010
• Issued: 03/21/2017
• Est. Priority Date: 10/22/2010
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Srinivasan, Kiran, Bisson, Timothy C.
• Primary Examiner(s): Murray, Daniel C
• Application Number:
  US12/910,535
• Publication Number:
  US 20120102135A1
• Time in Patent Office: 2,342 Days
• Field of Search: 709/213, 709/204, 714/4.11, 714/13
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1438 Restarting or rejuvenating
• CPC Class Codes: G06F 11/1484 involving virtual machines
• CPC Class Codes: G06F 11/2043 where the redundant components share a common memory address space
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects
• CPC Class Codes: H04L 67/1027 Persistence of sessions during load balancing

Abstract

The disclosed technique uses virtual machines in solving a problem of persistent state for storage protocols. The technique provides for seamless, persistent, storage protocol session state management on a server, for higher availability. A first virtual server is operated in an active role in a host system to serve a client, by using a stateful protocol between the first virtual server and the client. A second, substantially identical virtual server is maintained in a passive role. In response to a predetermined event, the second virtual server takes over for the first virtual server, while preserving state for a pending client request sent to the first virtual server in the stateful protocol. The method can further include causing the second virtual server to respond to the request before a timeout which is specific to the stateful protocol can occur.

Claims

1. A method comprising:

operating a first virtual server in an active role in a host processing system to serve a client, by using a stateful protocol between the first virtual server and the client;

associating a second virtual server in a passive role with the first virtual server;

in response to a predetermined event, causing the second virtual server to take over the active role from the first virtual server, while preserving state for a pending client request sent to the first virtual server in the stateful protocol; and

sharing state between the first and second virtual servers prior to the predetermined event, by sharing a designated region of memory between the first and second virtual servers, wherein prior to the predetermined event, the first virtual server has read-write permission to access the region of memory and the second virtual server has read-only permission to access the region of memory.

2. A method as recited in claim 1, further comprising:

causing the second virtual server, acting in the active role, to respond to the pending client request so that a timeout which is specific to said stateful protocol does not occur.

3. A method as recited in claim 1, further comprising:

maintaining state for client requests, conforming to the stateful protocol, in the designated region of the memory, prior to the predetermined event.

4. A method as recited in claim 3, wherein causing the second virtual server to take over the active role from the first virtual server, while preserving state for a pending client request sent to the first virtual server in the stateful protocol, comprises:

changing permission of the second virtual server to access the region of memory from read-only to read-write.

5. A method as recited in claim 4, wherein causing the second virtual server to take over the active role from the first virtual server, while preserving state for a pending client request sent to the first virtual server using the stateful protocol, further comprises:

assigning to the second virtual server a network address associated with the first virtual server.

6. A method as recited in claim 1, wherein causing the second virtual server to take over the active role from the first virtual server comprises:

updating a host memory page table of the host processing system; and

updating a guest memory page table of the second virtual server.

7. A method as recited in claim 1, wherein the first and second virtual servers are virtual network storage servers.

8. A method comprising:

operating a first virtual machine in a host processing system as an active virtual storage server;

maintaining per-client session state for pending, received client requests, which conform to a stateful network storage protocol, in a designated region of a memory;

initializing a second virtual machine in the host processing system to be a passive virtual storage server partnered with the first virtual machine;

sharing said state for pending, received client requests between the first and second virtual machines, by sharing the designated region of the memory between the first and second virtual machines;

in response to a failure event associated with the first virtual machine, causing the second virtual machine to take over for the first virtual machine as active virtual storage server, without losing state for a pending client request sent to the first virtual machine in the stateful network storage protocol, including modifying a permission of the second virtual machine to access the designated region of the memory; and

causing the second virtual machine, acting as active virtual storage server, to respond to the pending client request before a response timeout which is specific to the stateful protocol can occur.

9. A method as recited in claim 8, wherein prior to the failure event the first virtual server has read-write permission to access the region of memory and the second virtual server has read-only permission to access the region of memory;

and

wherein modifying a permission of the second virtual machine to access the shared designated region of memory comprises changing permission of the second virtual machine to access the designated region of memory from read-only to read-write.

10. A method as recited in claim 9, wherein causing the second virtual server to take over the active role from the first virtual server, while preserving state for a pending client request sent to the first virtual server in the stateful protocol, comprises:

updating a host memory page table of the host processing system; and

updating a guest memory page table of the second virtual server.

11. A method as recited in claim 10, wherein causing the second virtual machine to take over for the first virtual machine as active virtual storage server further comprises:

assigning to the second virtual storage server a network address associated with the first virtual storage server.

12. A system comprising:

a physical host processing system;

a first virtual machine which operates in the physical host processing system as an active server to serve a client over a network, by using a stateful protocol between the first virtual machine and the client;

a second virtual machine configured as a passive partner of the first virtual machine, wherein the second virtual machine is pre-initialized and maintained in a state that passively mirrors a state of the active server, the second virtual machine configured to respond to a predetermined event by taking over as the active server for the first virtual machine while preserving state for a pending client request sent to the first virtual machine in the stateful protocol; and

a memory that includes a designated region shared by the first virtual machine and the second virtual machine, wherein prior to the predetermined event, the first virtual machine has read-write permission to access the region of memory and the second virtual machine has read-only permission to access the region of memory.

13. A system as recited in claim 12, wherein the first and second virtual machines operate as network storage servers.

14. A system as recited in claim 12, wherein the second virtual machine, after taking over as the active server, responds to the pending client request before a response timeout which is specific to the stateful protocol can occur.

15. A system as recited in claim 12, wherein the active server maintains state for client requests in the designated region of the memory.

16. A system as recited in claim 12, further comprising:

a hypervisor through which the first virtual machine and the second virtual machine access the memory.

17. A system as recited in claim 16, wherein the system is configured to change permission of the second virtual machine to access the region of memory from read-only to read-write as part of a takeover by the second virtual machine for the first virtual machine.

18. A system as recited in claim 17, wherein the system is further configured to update a host memory page table of the physical host processing system and a guest memory page table of the second virtual machine as part of the takeover by the second virtual machine for the first virtual machine.

19. A system as recited in claim 18, wherein the system is further configured to assign to the second virtual machine a network address associated with the first virtual machine as part of the takeover by the second virtual machine for the first virtual machine.

20. A storage server comprising:

a network interface through which to receive client requests over a network, the client requests conforming to a stateful network storage protocol;

a non-volatile mass storage facility to store data in response to one or more of the client requests;

a processor, coupled to the network interface and the non-volatile mass storage facility;

a memory; and

a storage device coupled to the processor and storing code which, when executed, causes operations to be performed in the storage server, the operations including

operating a first virtual server in the storage server in an active role to serve a client, by using the stateful network storage protocol between the first virtual server and the client;

associating a second virtual server in a passive role with the first virtual server; and

in response to a predetermined event, causing the second virtual server to take over the active role from the first virtual server, without losing state for a pending client request sent to the first virtual server in the stateful network storage protocol, wherein the operations further comprise sharing state between the first and second virtual servers prior to the predetermined event, by sharing a designated region of the memory between the first and second virtual servers, wherein prior to the predetermined event, the first virtual server has read-write permission to access the region of memory and the second virtual server has read-only permission to access the region of memory.

21. A storage server as recited in claim 20, wherein the operations further comprise:

causing the second virtual server, having assumed the active role, to respond to the pending client request so that a timeout which is specific to the stateful protocol does not occur.

22. A storage server as recited in claim 20, wherein the operations further comprise:

maintaining state for client requests, conforming to the stateful protocol, in the designated region of the memory, prior to the predetermined event.

23. A storage server as recited in claim 22, wherein causing the second virtual server to take over the active role from the first virtual server, without losing state for a pending client request sent to the first virtual server in the stateful protocol, comprises:

changing permission of the second virtual server to access the region of memory from read-only to read-write.

24. A storage server as recited in claim 23, wherein causing the second virtual server to take over the active role from the first virtual server, without losing state for a pending client request sent to the first virtual server in the stateful protocol, further comprises:

assigning to the second virtual server a network address associated with the first virtual server.

25. A storage server as recited in claim 20, wherein causing the second virtual server to take over the active role from the first virtual server comprises:

updating a host memory page table of the host processing system; and

updating a guest memory page table of the second virtual server.