Virtualization and high availability of network connections
A system and method for connection failover. A plurality of virtual machines connect to a proxy using a protocol library. The proxy establishes a connection to a gateway. A user context including a conversation identifier is stored. The proxy continues a conversation with a second virtual machine based on the conversation identifier and user context if a first virtual machine fails when the conversation is in a ready for failover state.
1. A method comprising:
initiating a conversation between a first virtual machine included on a client and an application included on a server, the conversation to include a request/response pair, the first virtual machine and the application to execute the conversation via an Internet Connection Manager (ICM);
storing a conversation ID identifying the conversation between the first virtual machine and the application;
in response to a failure of the first virtual machine during the conversation, determining the failure of the first virtual machine occurred outside the request/response pair; and
in response to determining the failure of the first virtual machine occurred outside the request/response pair
maintaining the connection from the ICM to the application,
passing the conversation ID to a second virtual machine included on the client, and
establishing communication between a second virtual machine included on the client and the ICM to continue the conversation.
2. The method of claim 1 further comprising:
maintaining a user context including the conversation identifier from the first virtual machine in a persistent storage unit.
3. The method of claim 1 , wherein initiating the conversation comprises:
sending a message between the first virtual machine and the ICM using a shared memory protocol.
4. The method of claim 1 further comprising:
deleting the connection in response to determining the failure of the first virtual machine occurred within the request/response pair.
5. The method of claim 1 , wherein the first and second virtual machines communicate with the ICM via a memory pipe, and wherein the ICM and the server communicate via a socket.
6. The method of claim 1 , wherein the conversation ID is stored in shared memory.
7. A machine-accessible storage medium containing instructions that when executed cause a machine to:
initiate a conversation between a first virtual machine included on a client and an application included on a server, the conversation to include a request/response pair, the first virtual machine and the application to execute the conversation via an Internet Connection Manager (ICM);
store a conversation ID identifying the conversation between the first virtual machine and the application;
in response to a failure of the first virtual machine during the conversation, determine the failure of the first virtual machine occurred outside the request/response pair; and
in response to determining the failure of the first virtual machine occurred outside the request/response pair
maintain the connection from the ICM to the application,
pass the conversation ID to a second virtual machine included on the client, and
establish communication between a second virtual machine included on the client and the ICM to continue the conversation.
8. The machine accessible storage medium of claim 7 , further comprising instructions causing the machine to:
maintain a user context including the conversation identifier from the first virtual machine in a persistent storage unit.
9. The machine accessible storage medium of claim 7 , wherein the instructions causing the machine to initiate the conversation comprise instructions causing the machine to:
send a message between the first virtual machine and the ICM using a shared memory protocol.
10. The machine accessible storage medium of claim 7 , further comprising instructions causing the machine to:
delete the connection in response to determining the failure of the first virtual machine occurred within the request/response pair.
11. The machine accessible storage medium of claim 7 , wherein the first and second virtual machines communicate with the ICM via a memory pipe, and wherein the ICM and the server communicate via a socket.
12. The machine accessible storage medium of claim 7 , wherein the conversation ID is stored in shared memory.
13. A system comprising:
a memory
a server including an application;
a client device including a first virtual machine and a second virtual machine; and
an Internet Connection Manager (ICM) to
initiate a conversation between the first virtual machine and the application, the conversation to include a request/response pair,
store a conversation ID identifying the conversation between the first virtual machine and the application in memory,
in response to a failure of the first virtual machine during the conversation, determine the failure of the first virtual machine occurred outside the request/response pair, and
in response to determining the failure of the first virtual machine occurred outside the request/response pair
maintain the connection from the ICM to the application,
pass the conversation ID to the second virtual machine, and
establish communication between a second virtual machine included on the client and the ICM to continue the conversation.
14. The system of claim 13 , wherein the first and second virtual machines communicate with the ICM via a memory pipe, and wherein the ICM and the server communicate via a socket.
15. The system of claim 13 , wherein the conversation ID is stored in shared memory.
16. The system of claim 13 , the ICM to further:
delete the connection in response to determining the failure of the first virtual machine occurred within the request/response pair.