WO2007049388A1

WO2007049388A1 - Search intermediating apparatus, search intermediating method, decentralized search system, decentralizing apparatus, and decentralizing apparatus control method

Info

Publication number: WO2007049388A1
Application number: PCT/JP2006/315334
Authority: WO
Inventors: Chisato Numaoka
Original assignee: Sony Computer Entertainment Inc.
Priority date: 2005-10-24
Filing date: 2006-08-02
Publication date: 2007-05-03

Abstract

A search intermediating apparatus capable of requesting a computer, with which a reliable relationship is established, to do a part of a search processing. A search intermediating apparatus, which starts a search in accordance with a search request, accompanied by search information, from a client and which outputs a search result, includes a one-chip multi-core processor (32) comprising two or more processors that store respective IDs and secret keys. At least one of the two or more processors transmits, based on the secret key stored in that processor, at least a part of the foregoing search information related to the ID stored in that processor to a searching apparatus having an authenticated search engine, and receives a search result from the searching apparatus.

Description

Specification

Search mediation apparatus, search mediation method, distributed search system, distributed processing apparatus, and distributed processing apparatus control method

Technical field

The present invention relates to a search mediating device, a search mediating method, a distributed search system, a distributed processing device, and a control method for the distributed processing device, and more particularly to distributed processing such as search using a multiprocessor. Background art

[0002] In a search system, in order to prevent the processing performance from being reduced due to the concentration of search processing on one search server, load distribution using multiple search servers, Some processors access multiple search databases in parallel.

[0003] For example, Patent Document 1 below discloses a parallel processing system that searches a text base at high speed. As shown in FIG. 20 (a), this system includes a plurality of search servers 2 that perform search processing and a search management server 1 for managing the operations of the plurality of search servers. Server 1 divides all relevant information based on the text base to be searched and the relevant text base, and combines the divided parts of the text base and related information corresponding to the divided parts into a plurality of searches. Assign to some or all of Server 2. Each search server 2 performs an information search for the text-based divided portion allocated by the search management server 1 in parallel and independently.

[0004] Also, in Patent Document 2 below, as shown in FIG. 4B, a search request for data stored in a plurality of secondary storage devices 5 is sent to a plurality of servers operating on a plurality of processors 4. There is disclosed a database management device 3 that performs load balancing of IZO and CPU by processing in parallel using processes and enables a reduction in response time.

[0005] Further, in Patent Document 3 below, as shown in FIG. 5C, a plurality of data bases that are distributed and accessible by the client 6 through one or more search servers 7 are also disclosed. A method for efficiently searching for source 8 is disclosed. On the other hand, in the system disclosed in Patent Document 4 below, when the server 13 having the search engine 14 receives the search request from the WEB browser 12 of the client 11 as shown in FIG. Accordingly, the index search unit 15 searches the database, and the P2P search unit 16 registers the peer-to-peer search for the peer node 21 that can be reached from the server 13 in the registered root node 17. Run through route node 20. Then, after the server 13 receives the result of the peer-to-peer search, the search result compiler 18 merges the search result of the index search unit 15 and the search result of the P2P search unit 16, and returns the result to the client 11. A method is disclosed Patent Document 1: Japanese Patent Laid-Open No. 9-134364

Patent Document 2: Japanese Patent Laid-Open No. 9-305631

Patent Document 3: Japanese Patent Laid-Open No. 10-21250

Patent Document 4: U.S. Patent No. 6636854

Disclosure of the invention

Problems to be solved by the invention

[0007] However, in a distributed 'parallel search system so far, there is no reliable trust relationship between the computer that makes the search request and the computer that executes the search. A search request is sent to an infected server or a computer with poor performance, and as a result, unintentional access to data or the home page provided by the malicious computer or a long response from a server with poor performance There was a possibility of waiting over time. In particular, as shown in FIG. 21, when a peer-to-peer search is performed on a large number of peer nodes 21, such a problem does not occur.

[0008] The present invention has been made in view of the above problems, and one of its purposes is a search mediating apparatus capable of requesting a part of search processing to a computer with established trust relationship, search mediation A method and distributed search system are provided.

[0009] Another object of the present invention is to provide a distributed processing device capable of promptly confirming the existence of a trust relationship with a computer when processing such as search is distributed to a plurality of computers. It is to provide a method for controlling a distributed processing apparatus.

Means for solving the problem

[0010] In order to solve the above-described problem, a search mediating apparatus according to the present invention is a search mediating apparatus that starts a search according to a search request from a client accompanied by search target information and outputs a search result. And a multi-processor including two or more processors each storing a secret key and a search engine having a search engine that is authenticated based on at least one of the two or more processors and the secret key stored in the processor A search target information transmitting means for transmitting to the device at least a part of the search target information related to an ID stored in the processor; a search result receiving means for receiving a search result from the search device; It is characterized by providing.

[0011] The search mediating apparatus according to the present invention is a search mediating apparatus that starts a search in accordance with a search request from a client with search target information and outputs a search result, each storing an ID and a secret key. At least one search device comprising: a multiprocessor including two or more processors; and a search engine authenticated based on a secret key stored in each of at least two of the two or more processors. Search target information transmitting means for transmitting at least a part of the search target information related to the ID stored in the processor, and search result reception in which each of the at least two processors receives a search result from the search device. And search result integration means for integrating the search results received by each of the at least two processors. And

[0012] In addition, a search mediating apparatus according to another aspect of the present invention includes a search request one-chip multiprocessor that makes a search request to at least one search apparatus, and a database that indexes search results of the at least one search apparatus. And a search result integration one-chip multiprocessor integrated into the system.

[0013] Further, a search mediating apparatus according to still another aspect of the present invention is a search mediating apparatus that makes a search request to a plurality of computers connected via a network in accordance with a search request from a client. Determining means for determining whether each computer is a trusted computer or an untrusted computer based on the presence or absence of an authentication process with each computer; and a high reliability test for displaying an answer from the trusted computer. Search result display means, and low reliability search result display means for displaying an answer from the unreliable computer separately from the display by the high reliability search result display means.

[0014] Further, the distributed search system according to the present invention is a distributed search system having a plurality of search devices operating in a distributed environment, wherein at least one search request device, at least one search mediating device, and at least two search devices. The search intermediary device includes a multiprocessor including at least two or more processors, and at least one processor included in the multiprocessor is one of the at least two or more search devices. After establishing a trust relationship with at least one, the search requesting device performs a search according to the received search request.

[0015] Further, the distributed processing device according to the present invention is connected to a plurality of computers via a network, and a plurality of distributed processing devices that request each part of given processing from the computers. And a plurality of processors each including a memory for storing at least authentication information, and each of the plurality of processors is based on the authentication information stored in a memory provided in the processor. After performing the authentication process with at least one of the computers, a part of the given process is requested to the at least one computer.

Brief Description of Drawings

FIG. 1 is a hardware configuration diagram of a computer used as a search mediating apparatus according to an embodiment of the present invention.

2 is a configuration example of a sub processor provided in the computer shown in FIG.

FIG. 3 is a diagram showing an example of a search system.

FIG. 4 is a diagram illustrating a configuration example of a peer node.

[Figure 5] Parallel search · Integrated batch report type search processing protocol diagram.

[Figure 6] Parallel search · Parallel sequential report type search processing protocol diagram.

FIG. 7 is a diagram showing another example of the search system.

FIG. 8 is a protocol diagram of search processing applied to the system shown in FIG.

FIG. 9 is a flowchart of peer node selection / authentication 'registration processing. FIG. 10 is a diagram showing an example of a trust relationship between peer nodes included in the search system according to the present embodiment.

FIG. 11 is a flowchart of a piano selection “authentication” registration process when one sub-processor performs authentication with a plurality of peer nodes.

[FIG. 12] —A diagram showing a configuration example of a piano keyboard access table when one sub-processor performs authentication with a plurality of peer nodes.

FIG. 13 is a diagram showing an example of a search interface (user interface).

FIG. 14 is a flowchart showing an example of search processing.

FIG. 15 is a diagram showing another example of a trust relationship between peer nodes included in the search system according to the present embodiment.

FIG. 16 is a flowchart of peer node selection / authentication / registration processing when one sub-processor authenticates with a plurality of peer nodes and also makes a search request to a peer node having no trust relationship.

FIG. 17 is a diagram showing a configuration example of a peer node access table when one sub-processor authenticates with a plurality of peer nodes and also makes a search request to a peer node having no trust relationship.

[FIG. 18] —A diagram showing an example of a search interface when one sub-processor authenticates with a plurality of peer nodes and also makes a search request to a peer node having no trust relationship.

FIG. 19 is a diagram showing a device configuration example of a search mediating apparatus according to the present embodiment.

FIG. 20 is a diagram showing a conventional distributed / parallel search.

FIG. 21 is a diagram showing a conventional peer-to-peer search.

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a system configuration example of a computer 31 used as a search relay device using the one-chip multi-core processor 32 used in the present invention. The one-chip multicore processor 32 includes a main processor 39 having a processor core 40, sub-processors 33a to 33h having a processor core 34, an lZ interface 48, and a memory controller. A total of 11 processing elements (information processing means) called “La 46” are coupled to the ring bus 45 and are configured to operate in parallel. Furthermore, each processing element is uniquely identified (identified) on the ring bus 45 by an ID such as “SPE1”, for example, which element sends a request to which element via the ring bus 45. ! / Can be specified.

The memory controller 46 is coupled to the main memory 47 and mediates each processing element accessing the main memory 47. In FIG. 1, the main memory 47 is drawn outside the one-chip multi-core processor 32, but the main memory 47 may be included inside the one-chip multi-core processor 32. South to connect various IZO devices (network interface 50, nonvolatile memory 51, DVD driver 52, input interface 54 such as keyboard connected to USB port 53, etc.) to I / O interface 48 The main processor 39 and sub-processors can be connected by connecting a graphics card 55 with a bridge 49 and frame memory 56 and controlling the display output to the monitor 57, and an RF processing unit 59 that enables wireless communication using the antenna 61. 33 can control external devices and exchange data through communication with external devices. This system is connected to a data communication network via a network interface 50, and the main processor 42 and the sub processor 33 can communicate with various devices on the network independently of each other. The received signal acquired by the RF processing unit 59 is frequency-converted by the down converter 60, converted to digital data by the AD converter 58, and then supplied to the IO interface 48.

[0020] Each of the sub processors 33 includes a core 34, a local storage (LS) 35, a memory flow controller (MFC) 36 including a DMAC (Dict Memory Access Controller) 37, and a secure memory 38 including a secret key. It is included as a component. The secure memory 38 is storage means such as a RAM that is accessed exclusively by the core 34 (that is, only from the core 34 provided in the same subprocessor 33 as the secure memory 38). Secure memory can be part of local storage (LS) 35, as shown in Figure 2 (a), or dedicated hardware (core 34 force exclusive) as shown in Figure 2 (b). RAM accessed by Or other storage means). In this case, the dedicated hardware may be a non-volatile memory or a volatile memory. Furthermore, as shown in Fig. 2 (a) and (b), the local storage (LS) 35 has a public key 63 and a public key certificate 64 indicating that the public key 63 has been certified by a third party. In addition, a peer node access table 65 holding information (address, etc.) related to access to external peer nodes is held.

[0021] Since each sub-processor 33 has its own local storage 35, once the program is loaded into the local storage 35, the subsequent processing is performed without accessing the main memory 47 unless necessary. Can continue.

The main processor 39 includes a core 40, an L1 cache, an L2 cache, and an MFC 43 including a DMAC 44 as constituent elements. Generally, the operating system operates on the main processor 39, and a program that operates on each sub-processor 33 is determined based on the basic processing of the operating system. Needless to say, the program power that operates in the sub-processor 33 may be a program that forms part of the operating system in the normal concept (for example, a device driver or a part of the system program). The main processor 39 and the sub-processor 33 are instruction set architectures (ISAs) each having a different instruction set.

[0023] In the present embodiment, what is common to all the forces capable of various variations is a process of performing mutual authentication between a sub-processor 33 and a peer node having at least one search engine. This process is performed before the actual search process (search request to peer node). Here, the search engine is a search engine provided by Google or Yahoo, for example. Here, first, as a simple example of this process, the case where each sub-processor 73 uniquely establishes a trust relationship with one peer node will be described with reference to FIG. 3 and FIG.

As shown in FIG. 3, in the distributed search system 70, each of the sub processor 33 and the peer nodes 84a to 84h stores at least one secret key. Although not shown in Figure 3, there is a public key certificate authority on network 85. The sub processor 33 and each peer node 84 are provided with a public key certificate from a public key certificate authority based on their public keys 63. You In other words, based on the public key 63 of the sub-processor 33, the public key certificate authority issues a public key certificate (such as the public key 63 encrypted with the private key of the certificate authority), and the public key certificate is issued. A reply is sent to the sub processor 33, and the sub processor 33 holds the public key certificate. Similarly, based on the public key of each peer node 84, the public key certificate authority issues a public key certificate and returns the public key certificate to the peer node 84, and each peer node 84 receives the public key certificate. Holding. As public key authentication infrastructure, X.509 recommended by ITU-T, for example, can be adopted.

As shown in FIG. 9, when each sub processor 33 establishes a trust relationship with the peer node 84, the sub processor 33 first selects the external peer node 84 in S61. Although not shown, a peer node selection server exists on the network here. The piano selection server uniquely selects one peer node 84 based on the request of each sub processor 33, and the sub processor 33 communicates with the selected peer node 84 in the SSL handshake protocol in S62. Mutual authentication is performed using For example, the sub-processor 33 receives the public key certificate from the selected peer node 84, decrypts the public key certificate using the public key of the public key certificate authority, and receives information on the peer node 84 (peer Node 84's public key and its administrator's name). The piano card 84 is authenticated from this information. Similarly, the peer node 84 also receives the public key certificate from the sub processor 33, decrypts the public key certificate using the public key of the public key certificate authority, and obtains information on the sub processor 33 (sub processor 33). The public key and the name of the user). This information power also authenticates the sub-processor 33. Thus, if it is confirmed in step S63 that mutual authentication has succeeded, the sub-processor 33 will thereafter make the peer node 84 a trusted peer node unless a failure such as inaccessibility occurs. to decide. Therefore, in step S64, the sub processor 33 registers information related to the peer node 84 in the peer node access table 65 shown in FIG. On the other hand, if the mutual authentication fails in S63, the process returns to S61 again, and the sub processor 33 requests the peer node selection server to select the peer node again. If the mutual authentication is successful until the end (for example, until a predetermined time elapses), the sub-processor 33 determines that there is no reliable peer node. in this case, In the present embodiment, the sub-processor 33 does not participate (participate in) the search processing distributed processing. The above processing may be executed when a search request is made from a client, or may be executed at other appropriate timing such as when the computer 31 is started up.

[0026] (First embodiment)

Next, the first embodiment will be described with reference to FIG. The first embodiment can be classified into the following two types depending on how search results are returned to the search requester.

[0027] (1) Parallel search · Integrated batch report type

(2) Parallel search · Parallel sequential report type

[0028] Hereinafter, the parallel search / integrated batch report type, which is the first type, will be described with reference to FIG. 3, FIG. 4, and FIG.

FIG. 3 shows how the mutual authentication between the sub-processor 33 and the peer node 84 has already been successful by the process of FIG. Eight sub-processors 33 are associated with eight peer nodes 84a to 84h via a network 85 such as the Internet or a home network. For example, sub-processor 33a is associated with peer node 84a and sub-processor 33b is associated with peer node 84b. In the example of FIG. 3, the power shown by the eight sub-processors 33 is established with the external peer node 84, and there are at least one sub-processor 73 involved in the search process. Good.

[0030] The peer node 84 shown in FIG. 3 may have the same configuration as the computer 31, or FIG.

A general purpose or special purpose computer having a secure memory 88 in the main memory 87 accessed from the main processor 86 or a secure memory 88 independent of the main memory 87, as shown in (a) and (b). It may be. If the peer node 84 adopts the configuration shown in FIG. 4, the private key 62 is stored in the secure memory 88, the public key 63, the public key certificate 64 issued by a third party, and the A peer anode access table 65 is provided. If the computer 31 in FIG. 3 is a single processor piano, the main processor 42 is the main processor 86 in FIG. The main memory 47 in 3 can be configured to be the main memory 87 in FIG. That is, the main memory 47 has a storage area corresponding to the secure memory 88 as shown in FIG. 6A, stores the private key 62 therein, and further stores the public key 63 and the public key in another storage area. Remember key certificate 64. However, if the computer 31 is a peer node and the main processor 42 is not involved in the search process, the computer 31 does not need to store the piano access table 65.

Next, the distributed search process will be described step by step with reference to FIG. First, in step S 1, a search request (search request data) is input from the client computer 83 on the network 85 (see FIG. 3) or via the input interface 54 such as a keyboard connected to the computer 31. For example, the search request is accepted by one of the main processor 42 or the sub processor 33 (hereinafter referred to as main PE). Search requests include search data such as text data (keywords, natural language sentences, etc.), image (PICT etc.), video (AVC or MPEG4 etc.), sound (MP3 or AAC etc.) or voice. The search data is written on the shared memory on the main memory 47 so that it can be read by the main processor 42 and all sub-processors 33 in step S2.

Here, when the data is written in the main memory 47, the search data can be divided and allocated to each sub processor 33. For example, when N search keywords are given, the N search keywords are divided into the number of sub-processors 33, and a part of the search keywords is assigned to each sub-processor 33. The search keyword assigned to each sub processor 33 is stored in association with the ID of the sub processor 33 or information equivalent to the ID. More specifically, for example, assume that the following keyword set is specified.

Keyword = {distributed, system, parallel, computer, 2000}

Here, it is assumed that the search system calculates the logical sum of data including each keyword. If there are three sub-processors, SPU1, SPU2, and SPU3, the above keyword is divided into three sets such as {Distributed, System}, {Parallel, Computer}, and {2000}. Ask. Then, associate each subset with the ID of each SPU. Write to in-memory 47. Thereby, each sub processor can read a keyword group to be searched by the sub processor from the main memory 47 based on the ID assigned to the sub processor.

[0033] Further, in step S3, each search intermediary PE (at least one of the sub-processors 33 in FIG. 3 executes a search process using the assigned search keyword independently). On the other hand, the fact that the search data has been written on the main memory 47 is notified using means such as an interrupt or asynchronous message communication. In another embodiment, each search intermediary PE can monitor a specific memory location corresponding to the search intermediary PE and move to step S4 when new data is written.

[0034] In step S4, each search intermediary PE issues an access request to the main memory 47 in parallel, and reads the search data allocated to the search intermediary PE. As described above, when an ID is added to a search keyword, the search intermediary PE can read only the keyword with an ID that matches its own ID. In step S5, the search intermediary PE uses the keyword read in step S4 as search data, and sends its ID and public key certificate to the peer node on the network 85 having the search engine associated as described above. A search request is sent with 64 attached (step S5).

[0035] When the search is completed, the peer node returns the search result in step S6 with the common key certificate of the peer node attached to the search intermediary PE having the ID included in the request (step S6). . The search intermediary PE that received the search result reply writes the search result to the main memory 47 with its own ID attached (step S7), and writes the search result to the main memory 47 for the main processor 42. The completion is notified using an interrupt or asynchronous message communication (step S8).

[0036] After sending the search data write notification to the search intermediary PE in step S3, the main processor 42 sends the search results to the main memory 77 from all search intermediary PEs involved in the search process. Wait for notification of writing (step S10), and then, in step S9, request the main memory 77 to read the search results and obtain all the search results. After that, all the search results, for example, the number of keywords Sort by a certain standard, such as many, order, etc., and reassemble as a single search result data (step S11), and in step 12, send the data to the client computer 83 that issued the search request It is processed whether it is displayed on the user interface such as a display or a display connected to the computer 71. Further, the completion of the search process may be reported to the client computer 83 or the computer 71 (that is, itself) that has issued the search request.

Next, the second type parallel search / parallel sequential report type will be described with reference to FIG. 3 and FIG.

[0038] The search process will be described step by step with reference to FIG. Steps S21 to S27 correspond to steps S1 to S7, respectively, and thus the description thereof is omitted here. This type is characterized in that each search intermediary PE sends a search result report to the search requester independently. Therefore, in step S28, the search result report is reported directly (sequentially) in parallel to the search mediator PE client computer 83 or computer 71 in parallel with the operation of other search mediator PEs. It has been made. Simultaneously or in parallel with this step S28, the main PE is notified of the search result report (step S29). In step S23, the main PE sends a search data write notification to the search intermediary PE, and then the search results are written to the main memory 77 from all search intermediary PEs involved in the search process. After waiting for (step S30), the client computer 83 or the computer 71 reports completion of the search result (step S31).

[0039] (Second Embodiment)

Next, a second embodiment will be described with reference to FIG. In the second embodiment, simultaneously with distributed parallel search for peer nodes, search result integration and integration result report processing are also performed in parallel. In this sense, it can be said that it is a combination of the two types of the first embodiment. This example is implemented using a search result request 'result notification unit implemented using a computer 31-1 including a one-chip multicore processor 32-1 and a computer 31-2 including a one-chip multicore processor 32-2. Integrated search results · Integrated result reporting unit and two units Computer 91.

[0040] Search result integration • The integrated result reporting unit, of course, also searches for search results. • The result notification unit also integrates the results collected from the peer node 112 of the network 85. Search processing for databases stored in storage devices) and non-volatile memories such as flash memory)

[0041] As shown in the example of FIG. 3, in FIG. 7, at least one sub-processor 33-1 included in the one-chip multi-core processor 32-1 of the search result request 'result notification unit is connected to the network 85. A trust relationship has already been established by performing mutual authentication with at least one peer node 112 arranged in a distributed manner, and each sub-processor 33-1 that has received a search request has established a trust relationship. Requests a part of the search to the peer node 112 that is searching. For example, when the sub processor 93-la has established a trust relationship with the peer node 112a, it makes a search request to the peer node 112a.

Next, the search process will be described step by step with reference to FIG. First, in step S41, a search request from the client 83 on the network 85, or a search request power S through the user input interface of the computer 31, a search result request. Result notification unit main PE1 (for example, main processor 42-1) (Step S41). After this, the main PE1 issues a search request 1 to the main PE2 (for example, the main processor 42-2) of the search result integration / integration result reporting unit (step S42) and searches the main memory 47-1 Write the target data (step S43). The search data request / result notification unit search mediator PE (for example, at least one of the sub-processors 941) interrupts that the search data has been written to the main memory 47-1. Notification is made by asynchronous message communication (step S44). Here, the order of the relationship between step S42 and the pair of steps S43 and S44 does not matter. In step S45, the search intermediary PE that has received the search data write notification reads the search data (search keyword) associated with its ID in step S45, and has a peer node that is in a trust relationship (eg, peer node 112 in FIG. 9). Search request 2 is performed for at least one of the above (step S46). [0043] The main PE2 executes a search process for the local storage 110 in response to the search request 1, and writes the search result 1 as a result to the main memory 47-2 (step S47). Thereafter, the main PE 2 reports to the client 83 that the search result 1 has been written to the main memory 47-2 (step S48).

[0044] While the search result 2 from all the peer nodes 112 is returned, the main PE 2 is in a waiting state in step S54. During this time, when a search for a certain peer node is completed, a search result 2 is returned to the search intermediary PE that has issued a search request to the piano node (step S49). In response, the search intermediary PE sends the search result 2 to the search result integration-integrated result reporting unit's result writing PE (eg, one of sub-processors 33-2) associated with the search intermediary PE. Report (step S50).

[0045] Subsequently, the result writing PE calculates an index so that the search result data indicating the search result 2 can be indexed (step S51), and writes the indexed state into the main memory 47-2 (step S52). ) And the completion of writing search result 2 to main PE2 using an interrupt or asynchronous message communication (step S53) o Main PE2 has all peer node capabilities in step S54. When it is confirmed that the search result report has been received, a request to read all the search results 2 is issued to the memory 47-2, and the search result data indicating the search results 2 is read therefrom (step S55).

[0046] In step S56, a difference between search result 1 and search result 2 (a search result included in search result 2 but not included in search result 1) is calculated, and search data corresponding to the difference is calculated. Register based on the index in the database provided on the local storage 110 (nonvolatile memory such as HDD or flash memory).

[0047] After that, the main PE 2 sends the difference between the search results 1 and 2 to the client 83 (step S57) and reports the search completion to the main PE 1 of the search result request 'result notification unit (step S57). S58). Finally, data indicating that the requested search has been completed is transmitted from the main PE 1 to the client 83 (step S59).

[0048] In the above processing, the search power of main PE2 is assumed to be completed earlier than the search of all peer nodes 112. For example, the search result report (step S49) from other peer nodes 112 Later, writing of search result 1 by main PE2 (step S47) occurs. There is also the possibility of life. In this case, first, it waits for reports of search results 2 from all the peer nodes 112, calculates the difference between search results 1 and 2 and registers it in the database on the local storage 110. Thereafter, the result of merging search result 1 and search result 2 (data corresponding to the logical sum of search result 1 and search result 2) is reported to client 83 at the timing corresponding to step S57. In this case, step S48 becomes unnecessary.

[0049] If the search of main PE2 is completed after the search in all peer nodes is completed, main PE2 reports search result 2 first, although not shown in the figure. After that, after searching for the main PE2, step S56 is executed. In this case, step S48 is not necessary.

As described above, in the computer system described in the first embodiment in FIG. 3 and the second embodiment in FIG. 7, one sub processor of the mano core processor is connected via the network 85. After establishing a trust relationship with the peer computer by mutual authentication, the sub processor power also asks the peer computer to perform a search process, and each sub processor receives the search result. In addition, the computer itself including the sub-processor can also be called a peer node because it also has a function as a server that accepts other computer requests in addition to a function as a client that requests other peer computers to search. .

[0051] In the above embodiments, it has been assumed that one sub-processor establishes a trust relationship with only one peer node. However, one sub-processor may have a trust relationship with a plurality of peer nodes and make a search request. Also, as shown in FIG. 10, a tree network of peer nodes may be configured with the peer node 120 receiving the search request as the apex. For example, the peer node 120 that first receives a search request has established a trust relationship with the two peer nodes 121a and 121a, and the peer node 121a has established a trust relationship with the peer nodes 122a and 122b. The repetitive power Each piano keyboard may be a computer system using a one-chip multiprocessor as shown in Fig. 1! /.

[0052] In the trust relationship network as shown in Fig. 10, when the peer node 120 receives a search request from a client, the search request is transmitted downward (toward the end of the tree). The peer nodes 124a, 124b, 125, 126a, 126b, and 127 that are sown at the end of the trust relationship return the search results upward (in the root direction of the tree). Eventually, the search results of all the peer nodes are collected in the peer node 120 that first receives a search request from the client. A client (not shown) making a search request to the peer node 120 receives this search result and displays it to the user. If the peer node 120 itself is a search server having a user interface, the search result is displayed to the user via the user interface on the spot.

[0053] Now, in order to obtain an embodiment as shown in FIG. 10, the registration process of the peer node that has already established the trust relationship performed by each sub-processor already shown in the flowchart of FIG. 9 is more generally extended. There is a need to. This is shown in Figure 11. In the figure, first, in step S71, it is determined how many peer nodes are to be selected (how many peer nodes to try mutual authentication). Of course, if the number of selected peer nodes N is 1, the processing in FIG. 11 is essentially the same as the processing in FIG. N can also be infinite. However, in practice, the peer node access table 65 in FIG. 2 or 4 is generally finite.

[0054] First, in step S72, it is verified whether N is 0 or not, that is, whether or not authentication has been performed for N scheduled peer nodes. While N is not 0, the process proceeds to step S73 to check whether there is a peer node that can be selected. As a result, if there is a peer node that can be selected, N is decremented by 1 in step S74, and then the process proceeds to step S75 to select the peer node. If the peer node to be selected does not already exist, the registration process is terminated. In step S76, mutual authentication is performed with the peer node selected in step S75. For example, an SSL handshake protocol can be used. Finally, in step S77, it is confirmed whether the peer node authentication is successful. If the peer node authentication is successful, the process proceeds to step S78. Register in the note access table. If peer node authentication fails, return to step S72 before registering in the peer node access table. FIG. 12 shows an example of the peer node access table 65 of FIG. 2 or FIG. In this table, the peer node ID (here, the local ID seen from the computer that holds the table that is not the global ID), the public key certificate storage location that is the address in the local memory, and the registered are registered. The IP address of the peer node and a response indicating whether or not the registered peer node is currently accessible exist as items. In this embodiment, if the location on the other party's network is in some form, such as a power MAC address using an IP address. In the response column, “OK” is displayed! /, Indicates an accessible peer node, and “NO” indicates that access is not possible for some reason (eg network disconnection or computer power off). Show the peer node!

FIG. 13 shows an example of a user interface (search interface) for using this embodiment. This user interface is displayed on the monitor 57 of the computer 31 and includes a search start button 132, a search stop button 133, a search condition input unit 134, a search expression input unit 135, a search genre specification unit 136, and a search result display. Includes part 137. The search condition input unit 134 inputs the maximum search wait time as part of the search condition. In this example, there are three options: 1 minute, 10 minutes, and unlimited, with the check box of 1 minute marked. The search expression input unit 135 is an input form for inputting one or a plurality of search keywords. The search genre designation unit 136 designates a search genre (type of data to be searched), and web, image, sound, news, group, and directory are shown as options. For example, if a sound is selected, sound data is displayed on the search result display unit 137 as a search result.

FIG. 14 shows a main part of processing of the search application. This process is executed when searching by the computer 31 or the client 83. First, the search application (including the user interface program that displays the screen of FIG. 13) is started (step S81), and the search is performed. Wait for an instruction to start (step S82). When the search button 132 is pressed, the search time limit (for example, a fixed value) is stored in the memory by checking the conditions set in the search condition input unit 134 of the search interface in FIG. Step S83). Then check to see if a remote search server is specified. For example, it is checked whether the remote server location (location identification information) such as URL or IP address is set in a specified variable such as remote_server. If no search server is specified, search processing is started on the computer that is executing the search application. This search process continues until the time limit expires. If any search result is obtained within the search time limit, the search result is displayed on the search result display unit 137 of FIG. 13, for example. The above is expressed as an algorithm as follows. Here, the variable keywordjist is a variable indicating the keyword group input to the search expression input unit 135 in FIG.

[0058] [Table 1]

void search unt * keyword— 1 i st) i f (remote_server! = NULL) 〃 A remote search server is specified. (Request search processing from the search server.

search— request 1 (remote— server, keyword— 1 i s t);

l e l se l いない Not specified.

検索 Perform the search process on your own machine.

search— request 2 (keyword— l ist):

)

[0059] In the above description, all cases have been assumed only when a trust relationship is established between the sub processor requesting the search and the piano requesting the search. However, if you can clearly distinguish whether the search result is from a trusted peer node or a search result from an untrusted peer node without actually restricting it to such a high level, A method of using search results based on the user's responsibility is also conceivable. For example, in FIG. 15, the peer node that processes each part of the search request includes an untrusted peer node. This figure differs from Fig. 12 in that nodes in the tree network contain unreliable peer nodes. The peer nodes 142a, 144b, 145, and 146b are peer nodes that cannot establish a trust relationship. These are, for example, publicly issued by third parties A peer node that does not hold a key certificate. For this reason, for example, for the peer node 141a, the peer node 142b is an unreliable peer node, and therefore the search result from this peer node 142b is recognized as unreliable by 141a. The same is true between the peer node 142a and the peer node 144b and between the peer node 143a and the peer node 146b. It should be noted that if a node is unreliable, search results from all nodes connected to it (below it) are unreliable and recognized as search results. For example, even if the peer node 145 holds a public key certificate issued by a third party, if the peer node 145 comes through an untrusted peer node 142a, the peer node 141a Recognize it as an unreliable search result. This is because information may be tampered with on an untrusted node due to an untrusted node in between.

[0060] FIG. 16 is a flowchart showing a registration process of a peer node when one sub-processor authenticates with a plurality of peer nodes, establishes a trust relationship in a part of the sub-processor, and can include one. is there. The processing in this figure is based on the tree-structured network in Fig. 15, and is partially modified from the flowchart in Fig. 11. Specifically, what differs from FIG. 11 is the processing after step S77. In Fig. 11, if peer node authentication failed, no action was taken. However, in order to be able to issue a search request even to an untrusted peer node as shown in FIG. 15, FIG. 18 adds step S79 as a process after the peer node authentication has failed. Yes. In other words, when mutual authentication with a certain peer node succeeds, the peer node is registered in the peer node access table as a peer node having no trust relationship.

FIG. 17 shows an example of a peer node access table generated by the process of FIG. Compared with the peer node access table shown in FIG. 12, the peer node access table shown in FIG. Here, the reliability of the peer node ID 0004 is “none”. This indicates that the peer node that is determined to have failed in the peer node authentication in step S77, that is, is recognized as an unreliable peer node. Furthermore, the public key certificate storage location of this peer node ID 0004 is indicated as “NONE”. Untrusted peer nodes do not have public key certificates, so local memory This is because it cannot be stored.

FIG. 18 shows an example of a user interface (search interface) when a search request is processed by a tree-structure network including untrusted peer nodes as shown in FIG. In the user interface 141 shown in the figure, the screen area for displaying the search results is divided into a search result display unit 147 for a peer node having a trust relationship and a search result display unit 148 for a peer node having no trust relationship. Except for the above, it is the same as the user interface 131 shown in FIG.

FIG. 19 shows an example of the device configuration of the search mediating apparatus according to the present embodiment. In this configuration example, a monitor 160 such as a television receiver is used, and the main body of the search mediating device 151 displays a search result or accepts a search request using the monitor 160. The search intermediary device 151 includes an input interface such as a switch 154 for starting / stopping the device, a confirmation lamp (indicator) 155 indicating operation / non-operation, an IR light receiving unit 152 for receiving an infrared signal from a remote controller, and a keyboard 159. A connection interface 153 such as USB is provided. Here, the keyboard 159 is depicted as being wired and connected to the connection interface 153. Of course, if the connection interface 153 is a wireless communication interface, the keyboard 159 may be a wireless keyboard. It is possible to communicate with the connection interface 153 in accordance with a communication standard such as llaZbZg.

[0064] The remote controller 156 includes an infrared light emitting unit 157 and a keypad unit 158 including a plurality of numbers, characters, or direction instruction keys. Each key may be configured so that an input type such as numeric input, character input, or direction indication input can be selected with one key (input type selection key). In this case, one of the keys is an input type selection key, and further, the keypad unit is provided with a display unit for displaying the inputted character, and the display unit has a display unit for the currently selected input type ( Numbers, letters, or direction instructions) are displayed. The input character string is further displayed on the display unit. In another example, a plurality of lamps indicating the input type or one capable of displaying a plurality of colors with one lamp is provided instead of the display unit. And this displays the input type. It is more preferable that the keypad unit 158 has keys for character conversion and character selection. [0065] The search mediating apparatus 151 is configured as shown in FIG. 1, for example, and is connected to a monitor 160 (corresponding to the monitor 57 in FIG. 1) via a graphic card 55. The output generated by the processor of the device 151 is displayed. Specifically, as shown in FIG. 19, for example, a search interface 141 as shown in FIG. 18 is displayed. The search interface is operated using a user interface such as the remote control 156 or the keyboard 159. For example, by entering hiragana from the remote control 156 as “Hiragana” and inputting the key force of the keypad unit 158 in order, and converting it into kanji or katakana as appropriate, Can be typed in. Further, the input type may be set to “direction indication” by the remote controller 156, and the cursor may be moved between the buttons on the search interface 141 by instructing the direction of the key force of the keypad unit 158. . For example, by pressing the right arrow key provided on the keypad unit 158, the cursor position is set in order such as a search start button 132, a search stop button 133, a search condition input unit 134, a search expression input unit 135,. Can be changed.

Note that the present invention is not limited to the above embodiments. For example, in the above description, the example in which the present invention is applied to a search system using a network is taken up. However, the present invention is not limited to this, and various processes are distributed by a plurality of computers arranged on the network. It is also applicable to a system that For example, the computer 31 is configured as a distributed processing device, and is connected to a plurality of computers (such as peer node 84) via the network 85. Then, each computer is requested for each part of given processing such as numerical calculation and graphics processing. At this time, the distributed processing apparatus includes a one-chip multi-core processor 32 including a plurality of sub-processors 33 as in the configuration of FIG. 1, and each sub-processor 33 involved in the distributed processing has a local storage 35. Stores at least authentication information (at least a private key). Then, after causing each sub-processor 33 involved in the distributed processing to perform authentication processing with at least one of the plurality of computers based on authentication information stored in the secure memory 38, and after completing mutual authentication, Requests at least one computer for part of a given process. At this time, the request includes the ID of the sub-processor 33 that is the request source. The computer that has received the request returns the processing result including this ID to the distributed processing device that is the computer 31. Then, the sub-processor 33 that requested the distributed processing acquires a processing result including its own ID from various received data. Thereafter, for example, the main processor 42 or the like integrates the processing results in the sub processors 33 involved in the distributed processing, and outputs the integration result to the monitor 57 or the like, for example. In this way, it is possible to prevent the situation in which each sub processor 33 accesses the main memory 47 during mutual authentication and it takes a long time to read out the authentication information. Mutual authentication with computers is possible.

Claims

The scope of the claims

[1] A search mediating device that starts a search according to a search request from a client with search target information and outputs a search result.

A multiprocessor including two or more processors each storing an ID and a secret key, and at least one of the two or more processors, a search engine that is authenticated based on a secret key stored in the processor! A search target information transmitting means for transmitting at least a part of the search target information related to the ID stored in the processor to the search device comprising:

Search result receiving means for receiving a search result from the search device;

A search mediating apparatus comprising:

[2] A search mediating device that starts a search according to a search request from a client with search target information and outputs a search result.

A multiprocessor including two or more processors each storing an ID and a secret key; and at least two of the two or more processors each having a search engine authenticated based on a secret key stored in the processor Search target information transmitting means for transmitting at least a part of the search target information related to the ID stored in the processor to at least one search device;

Search result receiving means for receiving each of the at least two processors for the search device power search result; and

Search result integration means for integrating the search results received by each of the at least two processors;

A search mediating apparatus comprising:

[3] The search intermediary device according to claim 1 or 2, wherein the search target information is a character string, an image, a video, a sound, or a voice.

[4] The multiprocessor is a one-chip,

3. The search mediating apparatus according to claim 1 or 2, wherein

[5] One of the two or more processors includes an interface for receiving the search target information. The search mediating apparatus according to claim 1 or 2, wherein the search mediating apparatus is characterized in that:

[6] Each of the two or more processors transmits at least a part of the search target information in parallel.

3. The search mediating apparatus according to claim 1 or 2, wherein

[7] The apparatus further comprises a shared memory accessed by the two or more processors.

3. The search mediating apparatus according to claim 1 or 2, wherein

[8] The search mediating apparatus writes the search target information received by the interface in the shared memory in association with the ID of each processor.

7. The search mediating apparatus according to claim 6, wherein

[9] Each of the two or more processors reads at least a part of the search target information from the shared memory based on an ID of each processor.

9. The search mediating apparatus according to claim 8, wherein

[10] At least one of the two or more processors nC fe a public key certificate issued by a third party.

3. The search mediating apparatus according to claim 1 or 2, wherein

[11] At least one of the two or more processors includes a mutual authentication unit that performs mutual authentication with a search device including at least one search engine.

3. The search mediating apparatus according to claim 1 or 2, wherein

[12] The search mediating apparatus further includes search result integration means for integrating all search results received by the two or more processors.

The search intermediary device according to claim 1, characterized in that:

[13] The search result integration means is provided in one of the two or more processors.

13. The search mediating apparatus according to claim 12, characterized in that:

[14] The search mediating apparatus further includes search result answering means for answering the client with the integrated search result.

14. The search mediating apparatus according to claim 13, wherein

[15] The search result answering means is provided in one of the two or more processors.

15. The search mediating apparatus according to claim 14, wherein

[16] When at least one of the two or more processors receives a search result from the search device, the search result is sequentially answered to the client.

The search intermediary device according to claim 1, characterized in that:

[17] At least one of the two or more processors includes a search end detection unit that confirms that all the search device capabilities that have transmitted at least a part of the search target information have been received. Be

17. The search mediating apparatus according to claim 16, wherein

[18] A search request one-chip multiprocessor that makes a search request to at least one search device, a search result integration one-chip multiprocessor that indexes and integrates the search results of the at least one search device into a database,

A search mediating apparatus comprising:

[19] The search request one-chip multiprocessor is:

Two or more first processors identifiable by ID,

A search target information receiving interface for receiving search target information input;

Search target information transmitting means for transmitting at least a part of the ID of the first processor and at least a part of the search target information to at least a part of the two or more first processors provided with a search engine. When,

The first processor that has transmitted a part of the search target information includes search result receiving means for receiving a search result corresponding to the part from the search device that is a transmission destination.

19. The search mediating apparatus according to claim 18, wherein

[20] The search result integrated one-chip multiprocessor is:

Two or more second processors identifiable by ID,

Search result receiving means for receiving a search result received by the first processor from the first processor related to the second processor, at least partly of the two or more second processors;

Search result integration means for indexing and integrating the search results received by each of the second processors into the database; 20. The search mediating apparatus according to claim 19, further comprising:

[21] The system further comprises a non-volatile storage device in which search results are indexed and an integrated database is stored.

21. The search intermediary device according to claim 20, wherein

[22] The first processor transmits at least a part of the search target information to at least one of the second processors, and performs a search using a database stored in the nonvolatile storage device. To ask,

The search mediating apparatus according to claim 21, wherein the search mediating apparatus is characterized in that:

[23] The second processor that has received the request sequentially outputs search results when the search is completed.

23. A search mediating apparatus according to claim 22, wherein

24. The first processor, when a search result is received from the search device by the search result receiving means, transmits the search result to at least one of the second processors. The search intermediary device according to claim 23.

[25] The search result integrated one-chip multiprocessor includes a memory,

25. The range according to claim 24, wherein the second processor creates an index with respect to the received search result and writes information with an index added to the search result to the memory. Search mediation device.

[26] The search result integrated one-chip multiprocessor adds an index written in the memory when all search device capabilities requested by the search request one-chip multiprocessor are received. Search results, calculate the difference from the directly output database search result, and register the difference in the database.

26. The search mediating apparatus according to claim 25, characterized in that:

27. The search mediating apparatus according to claim 26, wherein the search result integrated one-chip processor outputs information indicating the difference.

[28] The non-volatile storage device is a hard disk storage device.

[29] The nonvolatile storage device is a flash memory.

[30] A search mediation method for starting a search in accordance with a search request from a client accompanied by search target information and outputting a search result.

At least one of the two or more processors included in the multiprocessor that respectively store the ID and the secret key. For a search device including a search engine that has been authenticated based on the secret key stored in the processor. A search target information transmission step of transmitting at least a part of the search target information related to an ID stored in the processor;

A search result receiving step for receiving a search result from the search device;

A search mediation method comprising:

[31] A search mediation method for starting a search according to a search request from a client accompanied by search target information and outputting a search result,

At least one of two or more processors each storing an ID and a secret key included in the multiprocessor, and at least one search device including a search engine that is authenticated based on the secret key stored in the processor A search target information transmission step of transmitting at least a part of the search target information related to an ID stored in the processor;

A search result receiving step in which each of the at least two processors receives the search device capability search result; and

A search result integrating step of integrating the search results received by each of the at least two processors;

A search mediation method comprising:

[32] A search mediating apparatus that makes a search request to a plurality of computers connected via a network in accordance with a search request from a client,

A means for judging whether each computer is a reliable computer based on the presence or absence of a predetermined authentication process with each computer; and a response to the search request for the reliable computer power is displayed. High reliability Sex search result display means;

A search intermediary device comprising: a low-reliability search result display unit that displays an answer to the search request for the unreliable computer power separately from the display by the high-reliability search result display unit.

[33] a determination means for determining whether the search means for performing the search process is in a force remote computer in the local computer;

When the search means is in a local computer, a local search program starting means for starting a local search program for performing search processing by the search means; and when the search means is in a remote computer, the local search program Remote search program starting means for starting a remote search program that causes the search means of the remote computer to perform search processing, which is different from a search program;

The search mediating apparatus according to claim 32, further comprising:

[34] A search time limit setting means for setting the search time limit is further provided.

34. The search mediating apparatus according to claim 33, wherein

[35] Time limit search result display means for displaying the search results received by the plurality of computer powers by the search time limit,

35. The search mediating apparatus according to claim 34, wherein

36. The search mediating apparatus according to claim 33, wherein the search means includes a multiprocessor including at least two processors.

[37] When at least one of the at least two processors succeeds in mutual authentication with at least one of the computers, the computer-powered search result is a reliable computer-powered answer. to decide,

37. The search mediating apparatus according to claim 36, wherein

[38] At least a part of each of the processors and the plurality of computers stores a private key, a public key, and a public key certificate issued by a third party, and the mutual authentication is performed using the public key certificate. By exchanging certificates and authenticating with both private keys, 38. The search mediating apparatus according to claim 37, characterized in that:

[39] A search mediation method for requesting a search to a plurality of computers connected via a network in accordance with a search request from a client,

A determination step of determining whether each computer is a computer that can be trusted by the presence or absence of a predetermined authentication process with each computer;

A high-reliability search result display step for displaying the reliable computer-powered answer;

A low-reliability search result display step for displaying an answer from the unreliable computer separately from the display in the high-reliability search result display step;

A search mediation method comprising:

[40] A distributed search system comprising multiple search devices operating in a distributed environment,

At least one search request device;

At least one search mediating device;

And at least two or more search devices,

The search mediating apparatus includes a multiprocessor including at least two or more processors, and at least one processor included in the multiprocessor establishes a trust relationship with at least one of the at least two or more search apparatuses; Search according to the search request received from the search request device;

A distributed search system characterized by that.

[41] In a distributed processing apparatus that is connected to a plurality of computers via a network and requests each part of given processing to each of the computers,

A multiprocessor including a plurality of processors, each processor including a memory for storing at least authentication information;

Each processor performs an authentication process with at least one of the plurality of computers based on authentication information stored in a memory included in the processor, and then performs a part of the given process with the at least one process. Request to other computers,

A distributed processing apparatus.

[42] At least the area of the memory in which the authentication information is stored is exclusively accessed by the processor provided with the memory.

A distributed processing apparatus.

[43] In a control method for a distributed processing apparatus that is connected to a plurality of computers via a network and requests each part of a given process from each computer,

The distributed processing apparatus includes a multiprocessor including a plurality of processors, and each of the processors includes a memory for storing at least authentication information;

After causing each processor to perform authentication processing with at least one of the plurality of computers based on authentication information stored in a memory provided in the processor, a part of the given processing is transferred to the at least one processor. Make one computer request

A control method for a distributed processing apparatus.