JP4658412B2 - Data sharing device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
INDUSTRIAL APPLICABILITY The present invention can be used in a wide range of digital contents such as storage devices, personal computers (PCs), servers, set-top boxes, and the like, which are handling digital data, which has been growing rapidly. .
[0002]
Recently, personal computers have become widespread, and high-speed LAN (Locl Area Network) 100 Mbps has become commonplace, and an Ethernet controller having a speed of 1 Gbps has become a price range that can be easily obtained. In addition, the network infrastructure has begun to be established at home, and a line speed of 64 Mbps with ISDN (Integrated Services Digital Network) to 1 Mbps with ADSL (Asymmetric Digital Subscriber Line), and recently about 100 Mbps with FTTH (Fiber To The Home) can be obtained at home. The time has come.
[0003]
However, despite the establishment of such a network infrastructure, the software configuration is still mainstream in the information network using the conventional client model. Recently, a peer-to-peer model network sharing technique that does not require a server has attracted attention, but there is a problem that stable sharing is not possible due to the dynamic model.
[0004]
In the present invention, stable network sharing is realized while being a peer-to-peer model.
[0005]
[Prior art]
As a conventional form of data sharing, there has been an information sharing system based on a client-server model (hereinafter referred to as a “crasaba model”). File sharing by the Windows network and NFS (Network File System) by UNIX are famous. However, generally, a server is required to have a high-performance CPU (Central Processing Unit) capable of withstanding a high load and I / O (Input-Output) specifications, and is expensive. In addition, management is complicated and requires a skilled administrator. If communication between the client and server was interrupted for some reason (worst crash), the client was shut down and helpless.
[0006]
In recent years, unlike conventional Kurazaba models, a dynamic network model (Peer-to-Peer) that does not require expensive servers and skilled administrators, and that can freely join and leave the network (PC power ON / OFF) ) Is attracting attention. Examples are Napster and Gnutella, which are famous for music distribution. However, since the peer-to-peer model is based on a dynamic communication model, communication with a transmission source peer (for example, a PC) is performed while searching for target data from a specific peer and starting download. Often suddenly stopped (for example, the router was turned off and the PC itself was shut down), and normal file copying was often not performed.
[0007]
FIG. 28 is an explanatory diagram of a conventional example. In FIG. 28, peer A, peer B, and peer C, which are devices such as PCs, are connected by a network. The peer A is provided with a control unit 11, a data transfer unit 12, a data search / determination unit 13, and a communication unit 16. The data search / determination unit 13 requests the peers B and C to search for data to be searched from the communication unit 16. When the data is found in the peer B, the data is downloaded from the peer B.
[0008]
In such peer-to-peer (hereinafter referred to as P2P), which has attracted attention in recent years, communication between peers is not guaranteed unlike the Kurazaba model. For example, communication with the transmission source peer B (for example, a PC) suddenly stops during downloading of data (such as MP3 which is a compression method standard) (the router is turned off, or the PC itself is often shut down).
[0009]
If the download was forcibly terminated in the middle, it was considered to rescue it by the following method.
[0010]
(1) Re-execute from the same peer (simple re-execution, re-execution when communication is restored)
(2) Search for another peer and re-execute (There may be no other peer with data, but there was also a backup)
Napster and Gnutella take the above methods (1) and (2). Napster is managed by the server only for data location. The entity (data) is at each peer. Gnutella is the location of data and the entity (data) is also each peer. It was complete P2P.
[0011]
[Problems to be solved by the invention]
The conventional device has the following problems.
[0012]
(1): If download is forcibly terminated in the middle, re-execution from the same peer in (1) above (simple re-execution, re-execution upon communication recovery) may not return the same peer to an immediately executable state. It was. For example, when a peer is shut down during download, or when a blue screen (an unrecoverable screen, etc.) appears, the personal computer cannot be started immediately. In this case, even if it waited for a certain specific time, it was not able to be performed again, and this method itself was not able to be performed.
[0013]
(2): In the above (2), searching for another peer and re-execution (there may be no other peer with data, but there was also a backup), it is possible to search for another peer, but not necessarily Since the target data is not guaranteed to exist in other peers, it could not be found. If it is particularly popular or frequently used data, it is likely to exist, but if not, it may exist only at a specific peer. In such a case, this method itself was not feasible.
[0014]
In the present invention, in view of the above-described problems, a preparation is made to ensure that a search can be performed. That is, a shared group (service group) for receiving services is created, and a replica is always created when new data is created in this group. And when it ends in the middle of the download after data search, it aims at improving the convenience of a user by providing a mechanism in which other peers automatically take over the service.
[0015]
[Means for Solving the Problems]
FIG. 1 is a diagram for explaining the principle of the present invention, FIG. 1 (a) is a description of a network, and FIG. 1 (b) is a description of a data sharing apparatus. In FIG. 1, 1 to 3 are devices (peers A to C), 10 is a data sharing device, 11a is control means, 12a is data transfer means, 13a is data search / determination means, 14a is replica registration / reference means, 15a Is a buffer monitoring / determination means.
[0016]
The present invention has the following means in order to solve the conventional problems.
[0017]
(1): A data sharing apparatus for configuring a service group for sharing data between devices connected via a network, and data transfer means 12a for transferring data with other devices, and registration of data replicas And replica registration / reference means 14a for referencing. When new data is created by a device in the service group, the replica registration / reference means 14a creates and registers a replica of the new data. For this reason, when replicas are created on multiple devices connected to the network, the data is made redundant, and after searching for data from a specific device, communication fails during download from the found device In addition, downloads can be resumed from other devices. In addition, even if a specific device crashes, data can be recovered from its own replica data in another device.
[0018]
(2): In the data sharing apparatus of (1), a buffer for temporarily buffering data transfer and a buffer monitoring / determination unit 15a for monitoring and determining the buffer are provided, and the buffer monitoring / determination unit 15a includes The buffer amount of the buffer during data transfer is monitored, and if the data transfer is delayed, the data transfer is resumed by switching to another device. For this reason, when the amount of the transfer buffer decreases to an amount that causes a stagnation in communication, it is notified to the next download source, and the download can be continued without stagnation.
[0019]
(3): In the data sharing device according to (1) or (2), the data transfer means 12a first measures the transfer rate at the time of data transfer, and downloads from a device with a high speed. Do. For this reason, download time can be shortened.
[0020]
(4): In the data sharing apparatus of (1), (2) or (3), before data transfer, grasp the number of times data can be copied to create a replica and determine whether transfer is possible. . For this reason, it is possible to prevent an increase in copying by limiting the number of times that copying is possible.
[0021]
(5): In the data sharing apparatus according to (1) or (2), after new data is created in the service group, the total size and number of replicas already existing in the devices connected to the network are referred to. Then, the replica creation destination is determined so that the replica sizes in each device are balanced. For this reason, it is possible to perform efficient data sharing by balancing the size of replicas in each device by preferentially selecting a device having a small replica size.
[0022]
DETAILED DESCRIPTION OF THE INVENTION
In the present invention, stable network sharing is realized by the following procedure.
[0023]
(1) A service group is dynamically formed (peer-to-peer based one-to-one communication is performed a plurality of times or multicast (one-to-many) communication).
[0024]
{Circle around (2)} When new data is created somewhere on the peer, replicas are created on multiple peers (guaranteed to exist at other peers).
[0025]
(3) Data search is performed within the service group and downloaded from the peer with the fastest communication speed.
[0026]
(4) If the download is interrupted, it automatically restarts from the peer with the next highest communication speed in the middle of the data.
[0027]
(5) When downloading data, the data flow is monitored once through a buffer on the network.
[0028]
Next, a peer-to-peer data sharing method for realizing the above will be described in detail.
[0029]
(1): Description of entire block
FIG. 2 is an overall block diagram, and FIG. 2A is a diagram illustrating the configuration of the entire P2P network. 2A, peers A, Peer B, Peer C, Peer D, and Peer E, which are devices such as PCs, are connected by a network. Here, peer A is a data transfer request source, and peer D is a data download source having a replica. Further, the peer E becomes a buffer between the download source and the download destination as a buffer at the time of download, and is provided for grasping the transfer status.
[0030]
FIG. 2B illustrates the configuration within each peer. In FIG. 2B, a control unit 11, a data transfer unit 12, a data search / determination unit 13, a replica registration / reference unit 14, a buffer monitoring / determination unit 15, and a communication unit 16 are provided in each peer. . The control unit 11 performs overall control. The data transfer unit 12 controls data transfer. The data search / determination unit 13 performs data search and search result determination. The replica registration / reference unit 14 performs registration and reference of a replica that is a replica of the replica. The buffer monitoring / determination unit 15 monitors and determines a transfer buffer that temporarily buffers data transfer. The communication unit 16 performs communication between peers.
[0031]
When a specific peer requests data transfer, a search request is issued to all peers from a specific peer (here, peer A). Among the hit peers, a specific peer (here, peer D) is selected. The selected peer transfers the data to the requesting peer via the buffer (here, peer E).
[0032]
In addition to the conventional data transfer unit 12 and data search / determination unit 13, a replica registration / reference unit 14 and a buffer monitoring / determination unit 15 are newly added. The replica is created by other peers when new data is created, and the peer E (buffer) is a P2P buffer that acts as a buffer between the download source and the download destination at the time of download. Is provided to do.
[0033]
(1) Description of data transfer unit
The data transfer unit 12 is a data transfer unit that manages download transfer of data hit in the search.
[0034]
FIG. 3 is an explanatory diagram of the data transfer unit. In FIG. 3, the data transfer unit 12 includes a data transfer control unit 21, a management unit 22, a generation management unit 23, and a transfer unit 24. The data transfer control unit 21 controls data transfer performed with other peers via the communication unit 16. The management unit 22 manages the generation management unit 23, the transfer unit 24, and the like. The generation management unit 23 manages the number of copies. The transfer unit 24 performs data transfer.
[0035]
(2) Explanation of data search / determination unit
The data search / determination unit 13 requests to search for data to be searched, and in response to the response returned from each peer, the data search / determination unit determines the peer to download from the time and the contents of the replica list. It is.
[0036]
FIG. 4 is an explanatory diagram of the data search / determination unit. In FIG. 4, the data search / determination unit 13 includes a data search / determination control unit 31, a management unit 32, a search result determination unit 33, and a search unit 34. The data search / determination control unit 31 controls data search and determination. The management unit 32 manages the search result determination unit 33 and the search unit 34. The search result determination unit 33 requests to search for data to be searched by the search unit 34, and in response to a response returned from each peer, looks at the time and the contents of the replica list (for example, the replica size held), and which peer From which to download. The search unit 34 requests each peer to search for data to be searched.
[0037]
(3) Explanation of replica registration / reference unit
Upon receiving a replica creation request, the replica registration / reference unit 14 is replica registration / reference means for creating a replica while managing three list files. These three lists include a replica list for managing the overall size of replicas, the number of replicas, a group list for managing service groups, a data copy generation, and a generation management list for managing original peers.
[0038]
FIG. 5 is an explanatory diagram of the replica registration / reference unit. 5, the replica registration / reference unit 14 includes a replica registration / reference control unit 41, a replica management unit 42, a replica search unit 43, a replica creation unit 44, and a list 45. In the list 45, a replica list 46, a group list 47, and a generation management list 48 are provided.
[0039]
The replica registration / reference control unit 41 controls replica registration / reference. The replica management unit 42 manages the replica search unit 43, the replica creation unit 44, and the list file 45. The replica search unit 43 searches for replicas. The replica creation unit 44 creates a replica. The list 45 has three lists stored in a storage medium, for example, a hard disk (HDD) 60. The replica list 46 manages the total size of replicas and the number of replicas. The group list 47 manages service groups. The generation management list 48 manages data copy generations and original peers.
[0040]
(4) Explanation of buffer monitoring / determination unit
The buffer monitoring / determination unit 15 is provided to confirm whether communication is performed without any delay, and checks whether the flow of the P2P buffer is smooth (whether it is not delayed or underflowed). Buffer monitoring / determination means comprising a P2P monitoring unit and a P2P buffer determining unit that switches the download source to the next peer according to the result.
[0041]
FIG. 6 is an explanatory diagram of the buffer monitoring / determination unit. 6, the buffer monitoring / determination unit 15 includes a buffer monitoring / determination control unit 51, a P2P buffer management unit 52, a P2P monitoring unit 53, and a P2P buffer determination unit 54.
[0042]
The buffer monitoring / determination control unit 51 performs monitoring / determination control of the P2P buffer. The P2P buffer management unit 52 manages the P2P monitoring unit 53 and the P2P buffer determination unit 54. The P2P monitoring unit 53 checks whether the flow of the P2P buffer (HDD 60) is smooth. The P2P buffer determination unit 54 switches the download source to the next peer based on the monitoring result of the P2P buffer (HDD 60).
[0043]
(5) Description of HDD contents
FIG. 7 is an explanatory diagram of the contents of the HDD. In FIG. 7, the content of the HDD 60 includes three types of data and a list, that is, a list file 61, a replica (data) 62, and a P2P buffer 63. The list file 61 has three list areas (replica list 46, group list 47, and generation management list 48) as shown in detail in FIG. The replica (data) 62 is an area for holding original data and replica data. The P2P buffer 63 is a transfer buffer area that temporarily buffers data transfer.
[0044]
(2): Explanation of service group change
(1): Explanation of service group
FIG. 8 is an explanatory diagram of a service group, and FIG. 8A is an explanatory diagram of a logical network. In FIG. 8A, a group from peer A to peer F is configured. Each peer A to peer F is connected by a logical network (LAN).
[0045]
FIG. 8B illustrates a physical network. In FIG. 8B, peer B is connected to peer A, peer C, peer D, and peer E, and peer D is connected to peer F. The service group is logical, and the physical network is related to a case where the line is interrupted. In FIG. 8B, when the line between the peer B and the peer D is interrupted, the connection with the peer F other than the peer D is also interrupted.
[0046]
(2): Explanation of service group change (addition)
FIG. 9 is an explanatory diagram of changing (adding) a service group, and FIG. 9A is an explanation when the service group is two. In FIG. 9A, when the service group is 2, the IP (Internet Protocol) address of peer B is “203.139.XXX.12” and the IP address of peer C is “203.139.XXX.10”. Peer B and peer C exchange their own IP addresses with each other when creating a service group. It is registered in the group list 47. For example, the peer B group list 47 holds the IP address “203.139.XXX.12” of peer C.
[0047]
FIG. 9B illustrates a case where peer E participates in a group of peer B and peer C. FIG. 9B is a diagram in which the peer E having the IP address “203.139.XXX.13” has joined the service group in FIG. 9A. When Peer E newly joins, if he / she knows the IP address of Peer B, he / she can join and inform Peer B of his / her intention to join. Get a group list from peer B. Peer B notifies peers participating in all other groups that peer E has newly joined. Here, only peer C. Accordingly, for example, the peer B group list 47 holds the IP addresses “203.139.XXX.12” and “203.139.XXX.13” of the peer C and the peer E.
[0048]
FIG. 10 is a service group participation process flowchart. FIG. 10 is a flowchart showing the situation of FIG. Hereinafter, a description will be given according to the processes S1 to S3 of FIG.
[0049]
S1: The peer E connected to the network notifies the participation request to the peer B already participating in the service group.
[0050]
S2: Peer B that has received the participation notification from peer E notifies the participating peer (here, peer C) of the service group.
[0051]
S3: The group list 47 is updated at each peer (peers B, C, E), and this process is terminated.
[0052]
FIG. 11 is a flowchart of a service group withdrawal process. Update in the same way as the participation in FIG. In the following, description will be given according to processing S11 to S12 of FIG.
[0053]
S11: A peer connected to the network (a peer who wants to leave) notifies the peer already participating in the service group of the withdrawal. When you receive a notice of withdrawal from the peer, the withdrawal is complete. (Alternatively, the notified peer can notify the participating peers of the service group.)
S12: Each peer in the service group updates the group list 47 and ends this process.
[0054]
(3): Explanation of the list held by each peer
FIG. 12 is an explanatory diagram of the list (1) held by each peer. In FIG. 12, peer B with IP address “203.139.XXX.10” is peer A with IP address “203.139.XXX.15”, peer C with IP address “203.139.XXX.12”, and IP address “203.139.XXX”. .16 "is connected to peer D. Each peer maintains a list 45, and each list 45 includes a replica list 46 and a group list 47. For example, the replica list 46 of the peer B holds the total size of replicas and the number of replicas. The replica list 46 is used as information for creating a replica. The group list 47 includes addresses (IP addresses “203.139.XXX.15”, “203.139.XXX.12”, “203.139.XXX”) of each peer (A, C, D) used to hold a group for creating a replica. .16 ”).
[0055]
FIG. 13 is an explanatory diagram of the list (2) held by each peer. In FIG. 13, peer B is connected to peer A, peer C, and peer D. Each peer holds a list 45, and the list 45 holds a generation management list 48 in addition to the replica list 46 and the group list 47 of FIG. For example, the generation management list 48 of the peer B is an original peer (A, B,...) When a replica is created. The data file name (a.mpg, b.mpg,...) And the number of copies (1, 2,...) For managing the copy are held. Even if you have your own original, the generation management list has the same structure. The original peer becomes its own peer (Peer B). When replicas are created in other peers, a copy creation end notification is received and the number of copies is incremented. Synchronize with the replica creator.
[0056]
(3): Explanation of replica creation and search
FIG. 14 is an explanatory diagram of creation and search of replicas, and FIG. 14 (a) is an explanation when new data is created by the peer B. FIG. In FIG. 14A, new data is created by peer B, and replicas are created by peer A and peer D. The creation destination is confirmed after the replica list is confirmed (for example, confirmation of a peer having a small overall size).
[0057]
FIG. 14B is an explanation of search and download. In FIG. 14B, search and download are performed. After receiving the search results, the download destination communicates the PING (ping command) reaction time or small data of about 1 Kbyte to check the transfer time, and downloads it from a peer with high speed. In FIG. 14B, the search source peer is E, which is downloaded from peer B.
[0058]
FIG. 15 is a replica creation processing flowchart (1). FIG. 15 is a flow when data is newly created at some peer. Does not manage copy generation. When new data is created, the group list is referred to and a replica list is obtained from each peer. As many replicas as the number of replicas N (optionally one or more) are selected so that the replica sizes are balanced. A replica creation request is made to the selected peer, and a replica is created at each peer. Then, various lists are updated at each peer. If some of the peers are powered down, there is no system problem, as only the response from the powered-down device is not returned. Hereinafter, a description will be given according to processing S21 to S27 of FIG.
[0059]
S21: Data is newly created at some peer in the service group, and the process proceeds to S22.
[0060]
S22: The peer that created the new data refers to the group list 47, and proceeds to the process S23.
[0061]
S23: The peer that has created new data receives the replica list 46 from each peer in the group list 47, and proceeds to the process S24.
[0062]
S24: The peer that created the new data selects N peers with a small overall size from the received replica list 46, and proceeds to processing S25.
[0063]
S25: The peer that has created new data requests the N selected peers to create a replica, and proceeds to processing S26.
[0064]
S26: The peer that has received the request for replica creation creates the replica, and proceeds to processing S27.
[0065]
S27: Each peer updates various lists and ends this process.
[0066]
FIG. 16 is a replica creation processing flowchart (2). FIG. 16 is a flow when data is newly created at some peer. Decide whether to create a replica by managing the copy generation. When new data is created, the group list is referred to and a replica list is obtained from each peer. As many replicas as the number of replicas N (optionally one or more) are selected so that the replica sizes are balanced. Request replica creation for the selected peer. However, if the number of copies is other than the specified number, creation of the replica is rejected. If the number of copies is within the specified number, each peer receives it and creates a replica. Update various lists at each peer. The number of copies is synchronized with each peer by performing replica creation sequentially. When creating a replica of its own data, it always receives an end notification and increments the number of copies of the original node. If the specified number of times is exceeded, replicas cannot be created. For example, even when the power of the peer that created the replica is dropped, the number of copies is written in the original generation management list, so that the number of copies does not increase more than the permitted number. If some of the peers are powered down, there is no system problem, as only the response from the powered-down device is not returned. Hereinafter, a description will be given according to processing S31 to S39 in FIG.
[0067]
S31: Data is newly created at any peer in the service group, and the process proceeds to S32.
[0068]
S32: The peer that created the new data refers to the group list 47, and proceeds to processing S33.
[0069]
S33: The peer that created the new data receives the replica list 46 from each peer in the group list 47, and proceeds to the process S34.
[0070]
S34: The peer that has created new data selects N peers with a small overall size from the received replica list 46, and proceeds to processing S35.
[0071]
S35: The peer that has created new data requests the N selected peers to create a replica, and proceeds to processing S36.
[0072]
S36: The replica creation requesting side (peer that created new data) checks the number of copies of the generation management list, and proceeds to processing S37.
[0073]
S37: The replica creation requesting side determines whether the number of copies is smaller than the permitted number. If it is determined that the number of copies is smaller than the permitted number, the process proceeds to step S38. If not, the replica creation is prohibited and the process is terminated.
[0074]
S38: The peer that has received the request for replica creation creates a replica, and proceeds to processing S39.
[0075]
S39: Each peer updates various lists and ends this process.
[0076]
FIG. 17 is a flowchart for creating both new data and a replica at another peer. FIG. 17 shows a case where newly created data is held not by its own peer but by another peer. For example, when the storage location of the peer A is full (even if there is still room), it is possible to realize the case where there is no own disk. Also, even if there is an own HDD, data can be created if the capacity of the own HDD is insufficient and other peers in the service group have capacity. Hereinafter, a description will be given according to processing S41 to S48 in FIG.
[0077]
S41: New data is created at peer A, and the process proceeds to step S42.
[0078]
S42: The peer A that has created new data moves to the processing S43 when its storage area is full (this processing block can be omitted).
[0079]
S43: The peer A that has created the new data refers to the group list 47, and proceeds to the process S44.
[0080]
S44: The peer A that has created new data receives the replica list 46 from each peer, and proceeds to processing S45.
[0081]
S45: Peer A that has created new data selects N peers with a small overall size from the received replica list 46, and proceeds to processing S46.
[0082]
S46: Peer A that has created new data requests the selected peer to create an original + replica and proceeds to process S47.
[0083]
S47: The peer requested to create the replica creates the replica and proceeds to processing S48 (one of the requested N peers holds the original).
[0084]
S48: Each peer updates various lists and ends this process.
[0085]
(4): Explanation when a certain peer crashes
FIG. 18 is an explanatory diagram when a certain peer crashes. FIG. 18 shows a case where the peer C indicated by x has crashed. If C crashes, it will recover data from the other peers (A, B, D).
[0086]
FIG. 19 is a flowchart when a certain peer crashes. FIG. 19 is a flowchart of data recovery from another peer when a certain peer crashes or data cannot be read. For example, when peer C crashes and the HDD is replaced and restarted, it tries to join the service group again. When participating, it is recognized that the IP address is the same or that the same peer has joined with a GUID (Global Unit ID) (in this case, the GUID also needs to be held in the group list 47). Here, you can choose to join from scratch or restore data. In the case of new, it is not different from normal new participation. When recovery is selected, data is collected from each peer and recovered. Data is sequentially copied back from the peer having the data of peer C from the generation management list of each peer, and the data is restored. Note that the GUID is a unique ID, for example, when there is an Ethernet card MAC address (the hardware address of the Ethernet board is one address all over the world), and other 128-bit random values are automatically generated. There is.
Hereinafter, a description will be given according to processing S51 to S58 of FIG.
[0087]
S51: If peer C crashes, the process proceeds to process S52.
[0088]
S52: Re-start up peer C (no data in HDD), and proceed to processing S53.
[0089]
S53: Peer C that has re-started up notifies the service group (the peer of the existing participant) of participation, and proceeds to processing S54.
[0090]
S54: The peer notified of participation in the service group refers to the GUID, and proceeds to process S55.
[0091]
S55: The peer that has been notified of participation in the service group recognizes the peer C that has been restarted to participate as an existing group member, and causes the peer C to determine whether to restore data. If it is determined that the data is to be restored as an existing group member, the process proceeds to step S56. If the data is not restored, the process is terminated.
[0092]
S56: The peer (or peer C) notified of participation in the service group notifies the peer of the service group, asks if it has the file of peer C, and proceeds to processing S57.
[0093]
S57: The peer (or peer C) notified of participation in the service group copies back to the peer C in order from each peer having the file of the peer C, and proceeds to processing S58.
[0094]
S58: The peer (or peer C) notified of participation in the service group determines whether all copying has been completed. If it is determined that all copies have been completed, the process ends. If the copy has not been completed, the process returns to step S57.
[0095]
(5): Explanation of data search method
When searching for the target data, each peer is checked to see if it exists. If there is data, the answer is returned. If not, there is no answer. In order to find a peer having a high communication speed, a response to PING is seen or dummy data having a small data size is downloaded and the communication speed is measured. It is efficient to download data from peers with fast transfer speeds. Two methods shown in FIGS. 21 and 22 can be used to determine whether or not the download has been interrupted. FIG. 20 is a data search processing flowchart. Hereinafter, a description will be given according to processing S61 to S70 of FIG.
[0096]
S61: The peer that performs the data search refers to the group list 47 and proceeds to the process S62.
[0097]
S62: The peer that performs the data search notifies the search condition (for example, data file name a.mpg), performs the search, and moves to the process S63.
[0098]
S63: When a peer that has performed a data search receives a search hit from N peers, the process proceeds to process S64, and if there is no hit, this process ends.
[0099]
S64: The peer that performs the data search receives data of about 1 Kbyte for the hit N peers, measures the communication speed (or sees the response speed time of the PING), and moves to process S65.
[0100]
S65: The peer that has performed the data search creates a download destination ranking list (in order of speed), and proceeds to processing S66.
[0101]
S66: The peer that has performed the data search opens the communication path for the top N pieces, and proceeds to step S67.
[0102]
S67: The peer that has performed the data search starts downloading data from a peer with a high transfer rate, and proceeds to step S68.
[0103]
S68: The download-side peer (or the peer having the 2P2 buffer or the next ranking peer) determines whether the download is normal. If it is determined that the download is normal, the process ends. If the download is not normal (stops halfway (timeout)), the process proceeds to step S69.
[0104]
S69: The download-side peer (or the peer with the 2P2 buffer or the peer with the next ranking) selects the next candidate (the peer with the next highest transfer rate), and proceeds to the process S70.
[0105]
S70: The download-side peer (or the peer with the 2P2 buffer or the peer with the next ranking) determines whether there is a next candidate selected. If it is determined that there is no next candidate, this process ends. If there is, the process returns to step S67.
[0106]
(1): Explanation when the download side determines timeout
The download normal end determination method (1) is a method of determining a timeout on the download side (determining that the download has stopped midway). The time when the communication was interrupted when the download stopped halfway was measured at a specific time, and that time passed (timeout). The download requester recognizes the timeout and performs the download from the next peer again. FIG. 21 is a flowchart of the method (1) for determining whether download is normally completed. Hereinafter, a description will be given according to processing S71 to S75 of FIG.
[0107]
S71: The peer on the download side starts downloading, and proceeds to process S72.
[0108]
S72: The download-side peer determines whether data transfer is continuing. If it is determined that the data transfer is continued, the process returns to the process S71. If the data transfer is not continued, the process proceeds to the process S73.
[0109]
S73: The download peer determines whether the data transfer is complete. If it is determined that the data transfer is completed, the process is terminated. If the data transfer is not completed, the process proceeds to S74.
[0110]
S74: The download-side peer determines whether a timeout has occurred. If it is determined that a timeout has occurred, the process proceeds to step S75. If not, the process returns to step S71.
[0111]
S75: The download-side peer makes a transfer start request to the next candidate peer, and ends this process.
[0112]
{Circle over (2)}: Explanation in the case of determining the normal end of download using the P2P buffer
(A) of the method (2) for determining the normal termination of download is that the peer having the P2P buffer monitors and notifies the next peer, and (b) is that the next candidate monitors and switches by itself. The P2P buffer serves as a communication buffer. The P2P buffer is somewhere on each peer. Monitors the buffer underflow (data supply stagnation) and switches to the next transfer source when an underflow is likely to occur. The P2P buffer will be somewhere on the peer. The P2P buffer may be monitored by a peer having a P2P buffer or a next download candidate peer.
[0113]
FIG. 22 is a flowchart of the method (2) for determining whether download is normally completed, and FIG. 22 (a) is an explanation when a peer having a P2P buffer monitors. In the following, description will be given in accordance with steps S81 to S84 in FIG.
[0114]
S81: The peer having the P2P buffer periodically monitors the amount of the P2P buffer, and proceeds to step S82.
[0115]
S82: The peer having the P2P buffer determines whether the data transfer is completed. If it is determined that the data transfer has been completed, the process ends. If not, the process proceeds to step S83.
[0116]
S83: The peer having the P2P buffer determines whether it is underflow. If it is determined that there is an underflow, the process proceeds to step S84. If not, the process returns to step S81.
[0117]
S84: The peer having the P2P buffer switches communication to the next candidate peer, and ends this process.
[0118]
FIG. 22B illustrates the case where the next candidate peer monitors and switches by itself. Hereinafter, a description will be given according to processing S91 to S94 in FIG.
[0119]
S91: The next candidate peer periodically monitors the amount of P2P buffer, and proceeds to step S92.
[0120]
S92: The next candidate peer determines whether the data transfer is completed. If it is determined that the data transfer has been completed, the process ends. If not, the process proceeds to step S93.
[0121]
S93: The next candidate peer determines whether it is underflow. If it is determined that there is an underflow, the process proceeds to step S94. If not, the process returns to step S91.
[0122]
S94: The next candidate peer switches communication to itself and ends this process.
[0123]
In this way, since the next candidate peer monitors, data can be sent immediately if it is determined that there is an underflow.
[0124]
(6): Download explanation
FIG. 23 is an explanatory diagram of downloading from peer A to peer D. In FIG. 23, peer A to peer F constituting the service group are connected by a network. A P2P buffer is actually placed at each peer. FIG. 23 shows a configuration including a P2P buffer (download from A to D). A state in which download from peer A to peer D is via the peer B P2P buffer is shown. The P2P buffer exists somewhere in one peer and actually does not look like the figure, but goes through some peer.
[0125]
FIG. 24 is an explanatory diagram of a download source switch using a P2P buffer. In FIG. 24, a P2P buffer 63 is located at peer B, indicating a download from peer A to peer D. Also, the buffer amount (shaded portion) is monitored by the peer B or the next candidate peer C.
[0126]
That is, corresponding to FIG. 22, peer A is a data supply source (download data supply source), and peer D is a side requesting download. That is, the data supply source to the P2P buffer 63 is the peer A, and the data consuming peer of the P2P buffer 63 is the peer D. The P2P buffer 63 shows an example in the peer B. The next candidate for download is peer C. The hatched portion is the amount of data. If the hatched portion disappears before the transfer is completed, the data has underflowed. Switch to the next candidate peer C before underflow occurs (before the buffer amount is below a certain amount). Underflow is monitored by the next candidate peer C (the flow in FIG. 22A) or the peer B having the P2P buffer (the flow in FIG. 22B).
[0127]
(7): Explanation of data communication
FIG. 25 is an explanatory diagram of a communication interface. FIG. 25 shows an example of a socket interface used in UNIX, which is communication between client and server. In the socket interface, communication is roughly divided into three phases. (1) Connection connection phase, (2) Data transfer phase, and (3) Connection release phase. In the connection connection phase (1), “socket ()” that creates an endpoint and returns a socket descriptor, “bind” that names the socket, connection completion “listen ()”, and the server waits for connection “accept ()” ”And“ connect () ”waiting for connection. The data transfer phase {circle over (2)} includes read “read ()” and write “write ()”. In the connection release phase {circle over (3)}, there is a close socket “closesocket ()”.
[0128]
In this embodiment, when communication is interrupted during downloading, the connection connection phase {circle over (1)} can be completed and the data transfer phase {circle over (2)} can be immediately transferred so that communication with other peers can be started immediately. Keep it like that.
[0129]
FIG. 26 is an explanatory diagram of communication in the case of a firewall. In FIG. 26, peers A to D and peer F are connected via the Internet, and peer E and peer D having local addresses are connected via the NAT function. In this example, downloading is performed from peer D to peer E via a NAT function such as a router. In this way, communication is possible when either the download source or the download destination is the firewall source. Not both cases. The NAT (The IP Network Address Translator) function recently used in routers is used. Even a personal computer (PC) that has only an IP address (local address) that cannot be used on the Internet can be connected to the Internet.
[0130]
In other words, the NAT function converts a global address that can be connected to the Internet (an external network) and a local address that can be used only within a firewall. Even if multiple generations of terminals are connected under the router having the NAT function, they can be connected with one global address.
[0131]
(8): Description of digital data sharing device
FIG. 27 is an explanatory diagram of a digital data sharing apparatus. In FIG. 27, a digital data sharing apparatus is a network sharing apparatus that performs data sharing between devices connected via a network. The digital data sharing apparatus is provided with a CPU block 71, a network block 72, a memory block 73, an interface block 74, and a large capacity storage block 75.
[0132]
The CPU block 71 controls the whole. The network block 72 is for communicating with other devices. The memory block 73 is for storing a control program. The interface block 74 receives a request from the user. The large-capacity storage block 75 holds a replica and a communication buffer (HDD, DVD, MO, etc.).
[0133]
In the digital data sharing apparatus, when new data is created, a replica creation request or request is accepted on a plurality of other devices (digital data sharing apparatus) connected to the network.
[0134]
(9): Explanation of program installation
Control unit 11a, control unit 11, data transfer unit 12a, data transfer unit 12, data search / determination unit 13a, data search / determination unit 13, replica registration / reference unit 14a, replica registration / reference unit 14, buffer monitoring / determination The means 15a, the buffer monitoring / determination unit 15, the communication unit 16, and the like can be configured by a program, which is executed by the main control unit (CPU) and stored in the main memory. This program is generally processed by a computer. This computer includes hardware such as an input device which is an input unit such as a main control unit, a main memory (memory block), a file device, a display device, and a keyboard.
[0135]
The program of the present invention is installed on this computer. In this installation, these programs are stored in a portable recording (storage) medium such as a floppy disk or a magneto-optical disk, and a drive device for accessing the recording medium provided in the computer is used. Alternatively, it is installed in a file device provided in the computer via a network such as a LAN. Then, the program steps necessary for processing are read from the file device into the main memory and executed by the main control unit. By installing this program in a computer, it is possible to easily provide a data sharing apparatus that can resume download from another device when communication is impossible during download.
[0136]
As described above, the present embodiment has the following effects.
[0137]
(1) A service group is dynamically formed (peer-to-peer based one-to-one communication is performed a plurality of times or multicast (one-to-many) communication).
[0138]
As a result, by forming a group for sharing data, it becomes possible to share a data storage medium other than itself. It can also have security to deny access from peers outside the group. It is possible to identify the person to be notified, and does not increase unnecessary traffic on the network.
[0139]
{Circle around (2)} When new data is created somewhere on the peer, replicas are created on multiple peers (guaranteed to exist at other peers).
[0140]
Thereby, redundancy can be given to data and it can be assured that the same data exists in a plurality of peers. Even if one peer cannot communicate, data can be downloaded if another peer can communicate. If a specific peer crashes, it is possible to recover its own HDD from its own replica data in another peer.
[0141]
(3) Data search is performed within the service group and downloaded from the peer with the fastest communication speed.
[0142]
Thereby, since it is guaranteed that there are target data in a plurality of peers, it is possible to select from which of the plurality of peers to download. The peer with the fastest communication speed can be selected, which may save data download time. Since replicas are guaranteed to exist at a plurality of peers, parallel downloading is possible by specifying a plurality of peers as download sources.
[0143]
(4) If the download is interrupted, it automatically restarts from the peer with the next highest communication speed in the middle of the data.
[0144]
As a result, it is guaranteed that a plurality of peers have the same data. Therefore, when communication of one peer is interrupted, it can be changed to another peer immediately. If the communication speed of all peers is measured when the first peer is selected, if the communication of the fastest peer is interrupted, the next fastest peer can be found immediately. If you remember how far the data has been downloaded by the data size, even if you move to another peer, you can resume from the middle, and the total download time will not increase, even though you have changed the download source peer. In a normal backup configuration, a system that downloads data from another peer is also conceivable. However, in the present invention, a buffer (P2P buffer) is provided in the middle of the communication path so that communication is performed via the buffer, and the buffer amount is monitored. When the buffer is low, it can be seamlessly transferred to other peers.
[0145]
(5) When downloading data, the data flow is monitored once through a buffer on the network.
[0146]
As a result, download stagnation can be quickly detected by realizing the download through the network buffer. Even when the download source peer becomes unable to continue communication for some reason, the download source can be switched without being notified to the download receiver (without a time lag). In other words, seamless cooperation is possible.
[0147]
[Additional notes are described below]
(Appendix 1) A data sharing apparatus for configuring a service group for sharing data between devices connected via a network,
Data transfer means for transferring data with other devices;
A replica registration / reference means for registering and referring to replicas of data;
A data sharing apparatus, wherein when new data is created by a device in the service group, the replica registration / reference means creates and registers a replica of the new data.
[0148]
(Supplementary note 2) The data sharing apparatus according to supplementary note 1, wherein the data transfer means switches to another device and resumes download when communication becomes impossible during the download.
[0149]
(Supplementary note 3) A buffer that temporarily buffers data transfer;
Buffer monitoring / determination means for monitoring and determining the buffer,
2. The data sharing apparatus according to claim 1, wherein the buffer monitoring / determination unit monitors the buffer amount of the buffer during data transfer, and when data transfer is delayed, switches to another device to restart data transfer.
[0150]
(Supplementary Note 4) The buffer monitoring / determination means for monitoring the buffer amount is provided in a next candidate device or a device having the buffer which is switched when the data transfer is delayed and resumes the data transfer. The data sharing apparatus according to attachment 3.
[0151]
(Supplementary note 5) Supplementary note 3 is characterized in that the registration location of the replica and the location of the buffer are placed somewhere in the device in the service group connected to the network based on a list held in each device. The data sharing apparatus described.
[0152]
(Supplementary Note 6) When receiving a notification to join or leave the service group from another device, all other participants are notified based on the group list and the group list is updated mutually. The data sharing apparatus according to any one of appendices 1 to 4, which is characterized by
[0153]
(Supplementary note 7) The data transfer means first measures how much the transfer speed is at the time of data transfer, and downloads from a device with a high speed. Data sharing device.
[0154]
(Supplementary note 8) The data sharing apparatus according to any one of supplementary notes 1 to 5, characterized in that, before data transfer, the number of times data can be copied to create a replica is ascertained to determine whether transfer is possible. .
[0155]
(Additional remark 9) After new data is created in the service group, the arrangement is such that the size of the replicas in each device is balanced by referring to the total size and number of replicas already existing in the devices connected to the network. The data sharing apparatus according to any one of appendices 1 to 5, wherein a replica creation destination is determined so as to satisfy the requirement.
[0156]
(Additional remark 10) The data sharing means for comprising the service group which shares data between the apparatuses connected by the network,
Data transfer means for transferring data with other devices;
When new data is created by a device in the service group, as a replica registration / reference means for creating and registering a replica of the new data, a program for causing the computer to function or a computer-readable record recording the program Medium.
[0157]
【The invention's effect】
As described above, the present invention has the following effects.
[0158]
(1): When new data is created by a device in the service group, the replica registration / reference means creates and registers a replica of the new data, so that replicas are created on a plurality of devices connected to the network. Create and provide data redundancy, and after searching for data from one particular device, if communication fails during download from the found device, download from another device can be resumed. Even if a device crashes, you can recover data from your own replica data on other devices.
[0159]
(2): The buffer monitoring / determination means monitors the buffer amount of the buffer that is currently transferring data. If the data transfer is delayed, it switches to another device and restarts the data transfer. And downloads can continue without delay.
[0160]
(3): The data transfer means 12a first measures how much the transfer speed is at the time of data transfer and downloads from a device with a high speed, so the download time can be shortened.
[0161]
(4): Before data transfer, in order to grasp the number of times data can be copied for creating a replica and determine whether it can be transferred, the number of times that data can be copied can be limited to prevent an increase in the number of copies. it can.
[0162]
(5): After new data is created in the service group, refer to the total size and number of replicas that already exist in the devices connected to the network, so that the replica sizes in each device are balanced. In order to determine the replica creation destination, it is possible to prioritize and select a replica having a small replica size in the device and balance the size of the replica in each device, thereby performing efficient data sharing.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating the principle of the present invention.
FIG. 2 is an overall block diagram in the embodiment.
FIG. 3 is an explanatory diagram of a data transfer unit in the embodiment.
FIG. 4 is an explanatory diagram of a data search / determination unit according to the embodiment.
FIG. 5 is an explanatory diagram of a replica registration / reference unit according to the embodiment.
FIG. 6 is an explanatory diagram of a buffer monitoring / determination unit according to the embodiment.
FIG. 7 is an explanatory diagram of the contents of the HDD in the embodiment.
FIG. 8 is an explanatory diagram of a service group in the embodiment.
FIG. 9 is an explanatory diagram of changing (adding) a service group in the embodiment.
FIG. 10 is a service group participation process flowchart according to the embodiment;
FIG. 11 is a service group withdrawal processing flowchart according to the embodiment;
FIG. 12 is an explanatory diagram of a list (1) held by each peer in the embodiment;
FIG. 13 is an explanatory diagram of a list (2) held by each peer in the embodiment.
FIG. 14 is an explanatory diagram of replica creation and search in the embodiment;
FIG. 15 is a replica creation process flowchart (1) in the embodiment;
FIG. 16 is a flowchart (2) of replica creation processing in the embodiment.
FIG. 17 is a flowchart for creating both new data and a replica by another peer in the embodiment;
FIG. 18 is an explanatory diagram when a certain peer in the embodiment crashes;
FIG. 19 is a flowchart when a certain peer in the embodiment crashes;
FIG. 20 is a flowchart of data search processing in the embodiment.
FIG. 21 is a flowchart of a download normal end determination method (1) in the embodiment;
FIG. 22 is a flowchart of a method (2) for determining a normal download end according to the embodiment;
FIG. 23 is an explanatory diagram of download from peer A to peer D in the embodiment;
FIG. 24 is an explanatory diagram of a download source switch using a P2P buffer in the embodiment;
FIG. 25 is an explanatory diagram of a communication interface in the embodiment;
FIG. 26 is an explanatory diagram of communication in the case of the firewall according to the embodiment.
FIG. 27 is an explanatory diagram of a digital data sharing apparatus according to an embodiment.
FIG. 28 is an explanatory diagram of a conventional example.
[Explanation of symbols]
1-3 Equipment (Peers A to C)
10 Data sharing device
11a Control means
12a Data transfer means
13a Data retrieval / determination means
14a Replica registration / reference means
15a Buffer monitoring / determination means

Claims (4)

A data sharing device for configuring a service group for sharing data between devices connected via a network,
Data transfer means for transferring data with other devices;
A replica registration / reference means for registering and referring to replicas of data;
When new data is created in a device in the service group, the replica registration / reference means creates a replica of the new data and registers it in a device other than the device that holds the new data in the service group When new data is created in the service group, the total size and number of replicas already existing in the devices connected to the network are referred to so that the replica sizes in each device are balanced. A data sharing apparatus, characterized in that a replica creation destination is determined so as to achieve a proper arrangement . A buffer that once buffers the data transfer;
Buffer monitoring / determination means for monitoring and determining the buffer,
The buffer monitoring / determination means monitors the buffer amount of the buffer during data transfer, and if the data transfer is delayed due to an underflow of the buffer, it switches to another device holding the same data and restarts the data transfer. The data sharing apparatus according to claim 1.   3. The data sharing apparatus according to claim 1, wherein the data transfer means first measures the transfer rate at the time of data transfer and performs download from a device having a high speed. Since data transfer is not possible when the number of times the specified number of data copies can be exceeded, it is possible to grasp the number of times data can be copied of the data for creating a replica and determine whether transfer is possible before data transfer. The data sharing apparatus according to claim 1.

Images