Method of improving the reliability of peer-to-peer network downloads

[0001] The present invention relates to peer-to-peer networks and more particularly to a method of improving the reliability of peer-to-peer network downloads.

[0002] In a peer-to-peer network a computer connected to the network acts as both a server and a client. In one example a person may store JPEG (Joint Photographic Experts Group) photograph files on his computer. Other computers on the network can see the JPEG photograph files on his computer and request that they be downloaded to their computer. On the other hand this person can see the JPEG photograph on other network user's computers and may request that one of the files be downloaded to his computer. This makes peer-to-peer networks very versatile. One problem that occurs with peer-to-peer networks is that errors can occur during a download. The user then has to start the downloading process over. Since the “server” is not a dedicated server, this problem is even more likely to occur. Some of the sources of errors (failures) are: power down by a user during a download; server IP (Internet Protocol) address is dynamically re-assigned by the ISP (Internet Service Provider); and the server is not able to handle the load of requests. Note that this list is not meant to be exclusive, just representative.

[0003] Thus there exists a need for a method of downloading files on a peer-to-peer network that increases the reliability and efficiency of the downloading process.

[0014] A method of improving the reliability of peer-to-peer network downloads includes the steps of initiating a search from a client on a peer-to-peer network. Next, a list of servers is downloaded that satisfies the search. At least one of the servers is selected from the list of servers. Next, one of a plurality of downloading systems is selected based on a predetermined criteria. A file is downloaded using one of the plurality of downloading systems. One of the downloading systems is a multiple concurrent download system. Using this system the client downloads two or more identical versions of the file simultaneously. As a result, a failure on one of the servers does not result in the file not being downloaded. Generally this download system is only used if the bandwidth connection of the client is at least two times the potential download speed of the server.

[0015] FIG. 1 is a block diagram of peer-to-peer network 20 in accordance with one embodiment of the invention. The peer-to-peer network 20 has a plurality of peers (computers) 22, 24, 26, 28 connected together by a network 30. Note that two peers 26, 28 may be directly connected to each other. In one embodiment, the peer-to-peer network may contain a central index server 32. If the network 20 contains a central index server, the server 32 indexes the files (programs) that are available to members of the network 20 and handles queries for those files. In one embodiment, the queries may be text queries. In another embodiment, the queries are a unique key. In the unique key embodiment, a hashing code is used to determine the unique key for each file. The indexing method is separate from whether the peer-to-peer network has a central index server or uses a broadcast query method.

[0016] FIG. 2 is a flow chart of the steps used in a method of improving the reliability of peer-to-peer network downloads in accordance with one embodiment of the invention. The process starts, step 40, by originating a search from a client on a peer-to-peer network at step 42. A search query is broadcast over the peer-to-peer network at step 44. A list of servers and a list of associated document names that satisfy the search query is received at the client at step 46. At least one of the servers is selected from the list of servers at step 48. At step 50, a downloading system is determined based on a predetermined criteria. At step 52, a file is downloaded which ends the process at step 54. Note that once a user selects the correct file for downloading, the system may select exact copies of that file for downloading. In another embodiment, the downloading system may select files having the same content but a different compression level for downloading.

[0017] In one embodiment, a unique key that identifies the file is entered as the search query. In another embodiment, the user may also receive the file size, a source node and unique key with the list of servers.

[0018] In one embodiment, the system measures a connection speed to the plurality of servers. The connection speed of the plurality of servers is compared to an available bandwidth at the client. This information is used to determine which download system is used. Next it is determined if an available bandwidth is less than a connection speed to two of the servers. When the available bandwidth is less than the connection speed to two of the servers, a serial concatenated download system is selected. A serial concatenated download system selects a first server (one of the list of servers) to start a download. If there is a failure at the first server, a second server is selected. The download is requested to start at a next byte after a last byte received. When the available bandwidth is not less than the connection speed to two of the servers, a multiple concurrent download system is selected. This system starts downloading from multiple servers (at least two servers) at the same time. In another embodiment a multiple concatenated download system is selected when the available bandwidth is not less than the connection speed to two of the servers. The multiple concatenated download system starts a first download at a first byte of the file for one of the at least two servers. Then a second download is started at a second byte of the file for a second of the at least two servers. The files are combined to create a complete file. As will be apparent to those skilled in the art the system can be expanded to n-servers.

[0019] FIG. 3 is a schematic diagram of a downloading system in accordance with one embodiment of the invention. This figure shows an example of a serial concatenated download system. A first server 60 starts downloading a file 62 to the client from the beginning of the file (BOF). When an error 64 occurs, a second server 66 begins downloading the exact same file to the client at the point the error occurred. This reduces the time to download the file since only the parts of the file not received are downloaded. The two (or more) parts of the file are then concatenated to form the complete file 68. Note that the system may be expanded to include n-servers, as will be apparent to those skilled in the art.

[0020] FIG. 4 is a schematic diagram of a downloading system in accordance with one embodiment of the invention. This figure shows a multiple concurrent download system. A first server 70 and a second server 72 concurrently start to download the file 74, 76 to the client. If an error occurs with one of the servers the other server continues to download the file to the client. This increases the reliability of the download. If one servers completes the download first the other download is canceled. This can be accomplished by having two processes running on the client machine that send a message to each other when one of them has finished downloading the file or this can be done with two threads of the same process. Note that this system can be expanded to n-servers as will be apparent to those skilled in the art.

[0021] FIG. 5 is a schematic diagram of a downloading system in accordance with one embodiment of the invention. This figure shows a multiple concatenated download system. A first server 80 starts downloading a file 82 at a first start byte 84 to the client. Note that the start of file 84 need not be the beginning of the file 86. A second server 88 starts downloading the file 82 at a second start byte 90 to the client. When various portions of the file can be combined to from a complete copy of the file, the download is terminated and the portions are concatenated to form the complete file 92. This can be accomplished by having two processes running on the client machine that send a message to each other when an overlapping byte is downloaded or this can be done with two threads of the same process. Note that this technique can also be expanded to n-servers.

[0022] FIG. 6 is an example of a list of servers 100 that satisfy a search query in accordance with one embodiment of the invention. The figure shows an example of the information a user might receive in response to a text query. The user would receive a list of servers 102, an associated file name (file path) 104, a file size 106 for each of the files. Note that the file sizes 106 vary by the encoding scheme used 108. A source node 110 may also be provided for the file. A source node 110 is the server (file) from which the file was copied. Finally a bandwidth (download speed) 112 is included that shows how fast a file can be downloaded from the server.

[0023] FIG. 7 is an example of a list of servers that satisfy a search query in accordance with one embodiment of the invention. FIG. 7 is similar to FIG. 6 except that the search was based on a unique key (hashing code). In this case the file sizes 106 and encoding schemes 108 are exactly the same for all the servers. This is because the hashing code that determines the unique key is derived from the complete file. As a result a file encoded with two different encoding systems will have two different unique keys.

[0024] FIG. 8 is a flow chart of the steps used in a method of improving the reliability of peer-to-peer network downloads in accordance with one embodiment of the invention. The process starts, step 150, by initiating a search from a client on a peer-to-peer network at step 152. A list of servers that satisfy the search is received at the client at step 154. At least one of the servers from the list of servers is selected at step 156. In one embodiment, a user selects the server. In another embodiment, the client selects one of the files based on a file size, encoding scheme, download speed or other criteria. One of the plurality of downloading systems is selected based on a predetermined criteria at step 158. At step 160 a file is downloaded using one of the plurality of downloading system which ends the process at step 162.

[0025] In one embodiment the step of selecting a download system includes the choice of: a multiple concurrent download system; a multiple concatenated download system; or a serial concatenated download system. One of the predetermined criteria for selecting between the downloading systems is a connection speed to one of the servers. In one embodiment, the connection speed is compared to an available bandwidth at the client.

[0026] In one embodiment, initiating the search includes entering a text string. In another embodiment, initiating the search includes entering a unique key. The search may be broadcast to the peer-to-peer network. In another embodiment, the search query is transmitted to a central server. The response to the query may include a document name, a file size, an available bandwidth at a server and a source node for the file.

[0027] FIG. 9 is a flow chart of the steps used in a method of improving the reliability of peer-to-peer network downloads in accordance with one embodiment of the invention. The process starts, step 170, by initiating a search from a first peer to the peer-to-peer network at step 172. A list of peer servers that meet a search query is received at step 174. One of the plurality of downloading systems is selected based on a predetermined criteria at step 176. At step 178 a file is downloaded using the one of the plurality of downloading system which ends the process at step 180. In one embodiment, the predetermined criteria includes determining a connection speed to each of the peer server on the list of peer servers. A subset of the list of peer servers is selected based on the connection speed. One method of determining the connection speed is to receive a test file from each of the peer servers. Another method of determining the connection speed is to determine an order of response from each of the servers. Yet another method of determining connection speed is to ping each of the servers.

[0028] When an available bandwidth at the client is less than two times a connection speed, a server is selected with the fastest connection speed. The download is started from the server with the fastest connection speed. If the server with the fastest connection speed has an error before the file is downloaded, a second server is selected. A last byte received is determined. A download starting from a next byte command is transmitted to a second server.

[0029] When an available bandwidth at the client is not less than two times a connection speed, a plurality of servers is selected from the list of servers. A plurality of simultaneous downloads is started from the plurality of servers. If the client has received a complete version of the file from one of the plurality of servers, the rest of the downloads are terminated. In another embodiment, a plurality of simultaneous offset downloads is started from the plurality of servers. When a complete file can be formed from the plurality of simultaneous offset downloads, the complete file is constructed.

[0030] FIGS. 10a-c are a flow chart of the steps used in a method of improving the reliability of peer-to-peer network downloads in accordance with one embodiment of the invention. The process starts, step 190, by initiation a search form a first peer to the peer-to-peer network at step 192. A list of peer servers, a plurality of associated file names, a plurality of file sizes, a plurality of bandwidths and a plurality of source nodes that meet a search query are received at step 194. A connection speed to each of the peer servers on the list of peer servers is determined at step 196. A subset of the list of peer servers is selected based on the connection speed at step 198. When an available bandwidth is less than two times the connection speed, a server with a fastest connection speed is selected at step 200. A download is started from the server with the fastest connection speed at step 202. Next it is determined if the server with the fastest connection speed had an error before the file was download at step 204. When the server with the fastest connection speed had an error before the file was downloaded, a second server is selected at step 206. A last byte that was received is determined at step 208. A download starting from a next byte command is transmitted to a second server at step 210. When an available bandwidth is not less than two times a connection speed, a plurality of servers are selected from the list of servers at step 212. A plurality of simultaneous downloads is started from the plurality of servers at step 214. Next it is determined if one of the plurality of servers has received a complete version of the file at step 216. When one of the plurality of servers has received a complete version of the file at step 218, the rest of the downloads are terminate which ends the process at step 220.

[0031] Thus there has been described a technique that increases the reliability, ease and efficiency of downloading files over a peer-to-peer network. The method is flexible and adapts for the best solution depending on available bandwidth, connection speeds, etc.

[0032] The methods described herein can be implemented as computer-readable instructions stored on a computer-readable storage medium that when executed by a computer will perform the methods described herein.

[0033] While the invention has been described in conjunction with specific embodiments thereof, it is evident that many alterations, modifications, and variations will be apparent to those skilled in the art in light of the foregoing description. Accordingly, it is intended to embrace all such alterations, modifications, and variations in the appended claims.