System and method for distributing and indexing computerized documents using independent agents

• US 5,963,944 A
• Filed: 12/30/1996
• Issued: 10/05/1999
• Est. Priority Date: 12/30/1996
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Adams, Robert
• Primary Examiner(s): Rinehart, Mark H.
• Assistant Examiner(s): Thompson, Marc D
• Application Number:
  US08/777,517
• Time in Patent Office: 1,009 Days
• Field of Search: 395/200.32, 395/200.47, 395/200.48, 395/200.49, 707/3, 707/8, 707/10, 707/102, 707/104, 709/202, 709/217, 709/218, 709/219
• US Class Current: 1/1
• CPC Class Codes: G06F 16/1834 implemented based on peer-to-peer networks, e.g. gnutella p2p communication protocols H04L67/104
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/1095 Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
• CPC Class Codes: H04L 69/329 in the application layer [OSI layer 7]
• CPC Class Codes: H04L 9/40 Network security protocols
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A system is provided in which autonomous agents manage the distribution of data and index information among the nodes of a computer network. The system comprises a network of computer nodes, each of which includes a data store and an agent interface for execution of the autonomous agents. The autonomous agents move independently among the network nodes, using the agent interface at each node they visit to execute their functions. Various types of agents are employed to implement different functions in the system. Index agents transfer index fields to and from the data stores of the network nodes according to a transfer criterion that is biased to aggregate index blocks into index files on one or more network nodes. Replication agents replicate index files on different nodes to make the system robust against node loss. Balance agents break large index files into smaller subfiles, to prevent overloading any nodes of the network, and query agents locate and retrieve index file entries that match a query criterion. The different agent types manage index and data information on the network nodes without need for a central organizing computer.

Claims

1. A system for managing an index comprising index files distributed over the nodes of a computer network, the system comprising:

an agent interface on each node of the network, each agent interface including an interface package to provide access to resources on the node;

a data store on each network node capable of storing one or more index blocks; and

a plurality of autonomous index agents, each index agent being capable of moving among the network nodes to transfer index blocks among the data stores of the network nodes according to a transfer criterion selected to aggregate index blocks of a type on one or more nodes of the network.

2. The system of claim 1, further comprising one or more autonomous query agents, each query agent being capable of moving among the network nodes to locate an index file having a specified type, compare the index blocks of the index file with a specified value, and copy one or more index blocks having the specified value.

3. The system of claim 1, further comprising one or more autonomous balance agents, each balance agent being capable of moving among the network nodes to locate index files, compare a size of the located index file with a selected maximum size, and transfer a portion of the index file to a new node when the size of the index file exceeds the selected maximum size.

4. The system of claim 1, further comprising one or more autonomous replicating agents, each replicating agent being capable of moving among the network nodes to monitor a number of replicas of an index file and generate a new replica of the index file when the monitored number is less than a selected minimum number of replicas.

5. The system of claim 1, wherein the agent interface includes memory locations to store direction information on index files and timing information on the one or more autonomous index agents.

21. The system of claim 1, wherein the transfer criterion biases the autonomous agents to retrieve index blocks of a given type from nodes that have relatively few such index blocks and deposit them at a node that has relatively many index blocks of the given type.

22. The system of claim 5, wherein an autonomous index agent uses the direction information stored in the memory location of a current node to determine which network node to visit next.

23. The system of claim 5, wherein an autonomous index agent uses the timing information stored in the memory location of a current node to determine whether to access the data store of the current network node.

6. A method implemented by each of a plurality of autonomous agents to distribute data among a plurality of networked nodes, the method comprising:

identifying an index block at a first node of the network;

retrieving the index block, when a retrieve criterion is met;

moving to a second node of the network; and

adding the index block to an index field at the second node, when a deposit criterion is met, wherein the retrieve and deposit criteria are selected to aggregate index blocks of a given type on a network node, when the criteria are applied at each network node visited by the autonomous agents.

24. The method of claim 6, wherein moving to a second node of the network comprises:

reading hint information in the first node of the network; and

moving to a second node of the network that is indicated by the hint information.

25. The method of claim 6, wherein identifying an index block at a first node comprises:

checking a time marker at the first node; and

identifying the index block at the first node if the time marker indicates the autonomous agent has not recently visited the first node.

26. The method of claim 25, wherein the autonomous agent moves to a third node if the time marker indicates that it has recently visited the first node.

7. A method for managing an index that is distributed over a network of nodes, each of which includes a data store and an agent interface for executing agents received through the network, the method comprising:

detecting an agent at the agent interface of a network node; and

executing the agent to;

evaluate an index block transfer criterion; and

transfer an index block between the agent and the data store with a probability related to a number of index blocks already in the data store.

8. The method of claim 7, wherein executing the agent to transfer an index block comprises;

executing the agent to transfer the index block to the data store when the agent includes the index block.

9. The method of claim 8, wherein executing the agent to transfer the index block to the data store comprises executing the agent to:

identify an index block type for the included index block;

determine a number of index blocks in the data store having the identified index block type; and

deposit the index block in the data store when the number of index blocks exceeds a statistically determined threshold.

10. The method of claim 7, wherein executing the agent to transfer an index block comprises executing the agent when the data store includes the index block, to:

determine a number of index blocks in the data store having a selected index block type; and

retrieve the index block from the data store when the number of index blocks is less than a statistically determined threshold.

11. The method of claim 7, wherein executing the agent further causes the agent to retrieve hint information on a next node to be visited and transfer itself to the indicated next node.

27. The method of claim 11, wherein retrieving hint information further comprises identifying a next node at random, if no destination node is associated with the index block, and moving to the randomly identified next node.

12. A method for managing an index that is distributed over a network of nodes, each of which includes a data store and an agent interface, the method comprising:

detecting an agent at the agent interface of a network node; and

executing the agent to evaluate an index block transfer criterion and transfer an index block between the agent and the data store when the index block transfer criterion is met; and

providing a hint to direct the agent to a next node indicated by the transferred index block.

13. The method of claim 12, wherein executing the agent comprises:

executing the agent to evaluate a deposit criterion when the agent includes an index block and to deposit the index block in the data store when the deposit criterion is met; and

executing the agent to evaluate a retrieve criterion when the agent does not include an index block and to retrieve an index block from the data store when the retrieve criterion is met.

14. The method of claim 13, wherein evaluating the deposit criterion comprises:

identifying an index block type for the included index block;

determining a number of index blocks in the data store having the identified index block type; and

depositing the index block in the data store when the number of index blocks exceeds a statistically determined threshold.

15. The method of claim 13, wherein evaluating a retrieval criterion comprises:

determining a number of index blocks in the data store having a selected index block type; and

retrieving the index block from the data store when the number of index blocks is less than a statistically determined threshold.

16. A method implemented by an independent agent to distribute index data over a plurality of network nodes, the method comprising:

determining a number of index blocks of a given type at a first node;

retrieving one or more of the index blocks when the number of index blocks is less than a threshold value; and

moving to a second node of the network.

17. The method of claim 16, wherein moving to the second node comprises:

reading a direction hint the first node; and

moving to the second node indicated by the direction hint.

18. The method of claim 16, further comprising:

determining a number of index blocks having the given index type at the second node; and

transferring the retrieved index block to the second node when the number exceeds a threshold value.

19. A method for managing an index that is distributed over a plurality of network nodes, the method comprising:

identifying an index block at a first node of the network;

transferring the index block to an agent, when a number of similar index blocks at the first node meets a retrieve criterion;

transmitting the agent to a second node of the network; and

adding the index block to an index field at the second node, when a number of similar index blocks at the second node meets a deposit criterion.

20. The method of claim 19, wherein transferring comprises:

executing the agent in an environment provided at the first node to compare the number of similar index blocks at the first node with a threshold value and retrieve the index block to the agent when the number is less than the threshold value.

28. A distributed index system comprising:

a network of nodes, each node having a data store capable of storing one or more blocks of index data; and

a plurality of independent agents, each agent being programmed to sample the nodes of the network and, at each network node it visits, to transfer a block of index data to or from the data store with a probability determined by a number of similar index blocks in the data store.

29. The distributed index system of claim 28, wherein an index agent holding a block of index data determines a number of similar index blocks in a node'"'"'s data store and transfers the index block to the data store when the number exceeds a threshold value.

30. The distributed index system of claim 29, wherein the threshold value is determined statistically.

31. The distributed index system of claim 28, wherein an index agent determines a number of index blocks of a given type in a node'"'"'s data store and retrieves one or more of the index blocks when the number is less than a threshold value.

32. The distributed system of claim 31, wherein the threshold value is determined statistically.

33. The distributed system of claim 28, wherein each independent agents samples the network nodes by identifying a direction hint at a node it is currently sampling and transferring itself to the node indicated by the direction hint.

34. The distributed system of claim 33, wherein the direction hint indicates a randomly selected node.

35. The distributed system of claim 33, wherein the direction hint indicates a node according to an index block transfer implemented by the agent at a node it is currently sampling.

36. A system comprising:

a plurality of network nodes, each node having a data store capable of storing index data; and

a plurality of independent agents, each agent being capable of moving among the network nodes to sample the stored index data and redistribute selected blocks of the sampled index data according to a statistical criterion.

37. The system of claim 36, wherein each agent identifies a number of index blocks of a given type that are stored at a node and applies the statistical criterion to transfer index blocks of the type between the agent and node.

38. The system of claim 37, wherein the statistical criterion is a randomly selected threshold value that is compared with the identified number of index blocks.

39. The system of claim 36, further including a query agent initialized to indicate a value for a given index block type, the query agent being capable of moving among the network nodes to locate index blocks of a given type having the indicated value.