Efficient application hosting in a distributed application execution system

• US 7,877,482 B1
• Filed: 04/01/2008
• Issued: 01/25/2011
• Est. Priority Date: 04/01/2008
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: McAlister, Jon P., Gibbs, Kevin A., Barrett, Ryan, Ashcraft, Kenneth
• Primary Examiner(s): Phillips; Hassan
• Assistant Examiner(s): EDWARDS, JAMES A
• Application Number:
  US12/060,798
• Time in Patent Office: 1,029 Days
• Field of Search: 709/226
• US Class Current: 709/226
• CPC Class Codes: G06F 9/5022 Mechanisms to release resources
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

In a distributed application execution system having an application master and a plurality of application servers, each application server includes one or more processors and memory storing one or more programs. The one of more programs include instructions for storing in non-volatile storage a plurality of applications distributed to the application server by the application master, for loading into volatile storage and executing a respective application in response to a received request, and for returning a result to the request. In addition, the one of more programs include instructions for conditionally retaining the respective application in volatile storage, for responding to a future request, when criteria, including at least predefined usage level criteria, are met by the respective application, and otherwise removing the respective application from volatile storage upon returning the result to the request.

Claims

1. An application server, for use in a system having an application master and a plurality of application servers, the application server comprising:

one or more processors;

memory storing one or more programs to be executed by the one or more processors, the memory including volatile storage and non-volatile storage, the one of more programs including instructions for;

storing in the non-volatile storage a plurality of applications distributed to the application server by the application master;

loading into volatile storage and executing a respective application in response to a received request, and returning a result to the request; and

conditionally retaining the respective application in volatile storage, for responding to a future request, when criteria, including at least predefined usage level criteria, are met by the respective application, and otherwise removing the respective application from volatile storage upon returning the result to the request.

2. The system of claim 1, wherein the application server includes instructions for retaining a plurality of respective applications in volatile storage that meet the predefined usage level criteria.

3. The system of claim 2, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

4. The system of claim 2, wherein the application server includes instructions for retaining a number, greater than one, of instances of a respective application in volatile storage in accordance with predefined caching criteria, wherein the predefined caching criteria include:

the number of instances of the respective application in volatile storage available for handling new requests; and

the number of instances of the respective application handling active requests and not available for handling new requests.

5. The system of claim 2, wherein the application server includes instructions for limiting the number of applications in volatile storage.

6. The system of claim 1, wherein the application server includes instructions for removing an instance of the respective application from volatile storage according to the predefined usage level criteria and a predefined order for removal of application instances from volatile storage.

7. The system of claim 6, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

8. The system of claim 6, wherein the predefined order for removal of application instances is to remove first a least-recently-used application instance.

9. The system of claim 6, wherein the predefined order for removal of application instances from volatile storage is based at least in part on one or more service quality levels associated with the applications for which application instances are loaded in volatile storage.

10. The system of claim 1, wherein the application server further includes instructions for removing the respective application from volatile storage according to error criteria, wherein the error criteria include:

number of errors encountered when executing the respective application in response to the receive request; and

type of error encountered.

11. The system of claim 1, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

12. The system of claim 1, wherein each application of the plurality of applications is responsive to http requests.

13. A method of operating an application server in a system having a plurality of application servers, the method comprising:

storing in non-volatile storage of the application server a plurality of applications distributed to the application server by an application master;

receiving a request from a source external to the application server;

loading into volatile storage and executing a respective application in response to the received request, and returning a result to the request; and

conditionally retaining the respective application in the volatile storage, for responding to a future request, when criteria, including at least predefined usage level criteria, are met by the respective application, and otherwise removing the respective application from volatile storage upon returning the result to the request.

14. The method of claim 13, further including retaining a plurality of respective applications in volatile storage that meet the predefined usage level criteria.

15. The method of claim 14, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

16. The method of claim 14, further including retaining a number, greater than one, of instances of a respective application in volatile storage in accordance with predefined caching criteria, wherein the predefined caching criteria include:

the number of instances of the respective application in volatile storage available for handling new requests; and

the number of instances of the respective application handling active requests and not available for handling new requests.

17. The method of claim 14, further including limiting the number of applications in volatile storage.

18. The method of claim 13, further including removing an instance of the respective application from volatile storage according to the predefined usage level criteria and a predefined order for removal of application instances from volatile storage.

19. The method of claim 18, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

20. The method of claim 18, wherein the predefined order for removal of application instances is to remove first a least-recently-used application instance.

21. The method of claim 18, wherein the predefined order for removal of application instances from volatile storage is based at least in part on one or more service quality levels associated with the applications for which application instances are loaded in volatile storage.

22. The method of claim 13, further including removing the respective application from volatile storage according to error criteria, wherein the error criteria include:

number of errors encountered when executing the respective application in response to the receive request; and

type of error encountered.

23. The method of claim 13, wherein the predefined usage level criteria are based on a number of requests processed by a respective application in a predefined period of time.

24. The method of claim 13, wherein each application of the plurality of applications is responsive to http requests.

25. An application execution system, comprising:

a plurality of application servers, each of the application servers including non-volatile storage for storing a plurality of applications; and

volatile storage and computational resources for executing applications in response to requests received by the application execution system;

an application master that stores a plurality of applications in a library for distribution among the application servers, the application master including instructions for;

distributing applications from the library to the application servers for storage in the non-volatile storage of the application servers;

monitoring the application servers to generate usage information for each of the applications in the library for a predefined period of time ending at a current time; and

removing previously distributed applications from the non-volatile storage of respective application servers in accordance with the usage information;

wherein the distributing includes distributing each application to a respective number of the application servers determined by the application master in accordance with the usage information;

each application server of the plurality of application servers including instructions for;

storing in non-volatile storage the applications distributed to the application server;

loading a respective application into volatile storage, executing the respective application in response to a received request, and returning a result to the request; and

conditionally retaining the respective application in volatile storage, for responding to a future request, when criteria, including at least predefined usage level criteria, are met by the respective application, and otherwise removing the respective application from volatile storage upon returning the result to the request.

26. The system of claim 25 wherein the usage information used by the application master includes:

CPU load of each application server;

CPU usage of each application on a respective application server; and

amount of non-volatile memory available on each application server.

27. The system of claim 25 wherein the application master distributes from the library each application in the library to at least three application servers of the plurality of application servers.

28. The system of claim 25 wherein the removing of previously distributed applications from the non-volatile storage of the application servers and the distributing of applications from the library to the application servers are rate-limited with respect to how many applications are removed and distributed per predefined period of time.

29. The system of claim 25 wherein the application master manages the load on a respective application server, in accordance with predefined load requirements, by distributing applications to and removing applications from the non-volatile storage of the respective application server.

30. The system of claim 25 wherein the application master generates distribution data identifying which of the applications are present in the non-volatile storage of each of the application servers;

and wherein the system includes one or more front end servers that route requests to the application servers in accordance with the distribution data.

31. The system of claim 30, wherein the one or more front end servers route requests to the application servers based on load data collected from each of the application servers;

and wherein the load data includes number of outstanding requests sent to each application server.

32. The system of claim 25 wherein each application of the plurality of applications is responsive to http requests.

33. A method of operating a plurality of application servers and an application master in an application execution system, the method comprising:

at the application master;

storing a plurality of applications in a library for distribution among the application servers;

distributing applications from the library to the application servers for storage in the non-volatile storage of the application servers;

monitoring the application servers to generate usage information for each of the applications in the library for a predefined period of time ending at a current time; and

removing previously distributed applications from the non-volatile storage of respective application servers in accordance with the usage information;

wherein the distributing includes distributing each application to a respective number of the application servers determined by the application master in accordance with the usage information; and

at each application server of the plurality of application servers;

storing in non-volatile storage the applications distributed to the application server;

loading a respective application into volatile storage, executing the respective application in response to a received request, and returning a result to the request; and

conditionally retaining the respective application in volatile storage, for responding to a future request, when criteria, including at least predefined usage level criteria, are met by the respective application, and otherwise removing the respective application from volatile storage upon returning the result to the request.

34. The method of claim 33, wherein the usage information used by the application master includes:

CPU load of each application server;

CPU usage of each application on a respective application server; and

amount of non-volatile memory available on each application server.

35. The method of claim 33, wherein the application master distributes from the library each application in the library to at least three application servers of the plurality of application servers.

36. The method of claim 33, wherein the removing of previously distributed applications from the non-volatile storage of the application servers and the distributing of applications from the library to the application servers are rate-limited with respect to how many applications are removed and distributed per predefined period of time.

37. The method of claim 33, wherein the application master manages a load on a respective application server, in accordance with predefined load requirements, by distributing applications to and removing applications from the non-volatile storage of the respective application server.

38. The method of claim 33, wherein the application master generates distribution data identifying which of the applications are present in the non-volatile storage of each of the application servers;

and wherein the application execution system includes one or more front end servers that route requests to the application servers in accordance with the distribution data.

39. The method of claim 38, wherein the one or more front end servers route requests to the application servers based on load data collected from each of the application servers;

and wherein the load data includes number of outstanding requests sent to each application server.

40. The method of claim 33, wherein each application of the plurality of applications is responsive to http requests.