WO2009123865A2

WO2009123865A2 - Rebuilding data on a dispersed storage network

Info

Publication number: WO2009123865A2
Application number: PCT/US2009/037768
Authority: WO
Inventors: Lynn Foster; Jason Resch; Ilya Volvoski; John Guigley; Greg Dhuse; Vance Thornton; Dusty Hendrickson; Zackary Mark
Original assignee: Cleversafe, Inc.
Priority date: 2008-03-31
Filing date: 2009-03-20
Publication date: 2009-10-08
Also published as: EP2260387B1; EP2260387A2; US20100161916A1; WO2009123865A3; EP2260387A4; US8560882B2; US20120290873A1; US20130275833A1; US8880799B2; US8713358B2; US8868969B2; US8464096B2; US20080183975A1; US20130275844A1

Abstract

A rebuilder application operates on a dispersed data storage grid and rebuilds stored data segments that have been compromised in some manner. The rebuilder application actively scans for compromised data segments, and is also notified during partially failed writes to the dispersed data storage network, and during reads from the dispersed data storage network when a data slice is detected that is compromised. Records are created for compromised data segments, and put into a rebuild list, which the rebuilder application processes.

Description

REBUILDING DATA ON A DISPERSED STORAGE NETWORK

Inventors: Vance Thornton, Greg Dhuse, Jason Resch, liya Volvovski, Dusty Hendrickson, John Quigley, Zachary Mark, Lynn Foster

Cross-Reference to Related Applications [001] This application is a continuation-in-part of United States Patent

Application Number 11/403,391 , entitled "SYSTEM FOR REBUILDING DISPERSED DATA," filed April 13, 2006, which is a continuation in part of United States Patent Application Number 11/241,555, entitled "DIGITAL DATA STORAGE SYSTEM," filed September 30, 2005, and this application is also a continuation-in-part of United States Patent Application Number 11/973,542, entitled "ENSURING DATA INTEGRITY ON A DISPERSED STORAGE GRID," filed October 9, 2007, all of which are hereby incorporated by reference in their entireties.

Field of the Invention

[002] The present invention relates generally to systems, apparatus, and methods for distributed data storage, and more particularly to systems, apparatus, and methods for distributed data storage using an information dispersal algorithm so that no one location will store an entire copy of stored data, and more particularly still to systems, apparatus, and methods for rebuilding data on a dispersed data storage network that has been damaged, compromised or has experienced failure during read and write operations. Description of the Prior Art

[003] Storing data in digital form is a well-known problem associated with all computer systems, and numerous solutions to this problem are known in the art. The simplest solution involves merely storing digital data in a single location, such as a punch film, hard drive, or FLASH memory device. However, storage of data in a single location is inherently unreiiable. The device storing the data can malfunction or be destroyed through natural disasters, such as a flood, or through a malicious act, such as arson. In addition, digital data is generally stored in a usable fife, such as a document that can be opened with the appropriate word processing software, or a financial ledger that can be opened with the appropriate spreadsheet software. Storing an entire usable file in a single location is also inherently insecure as a malicious hacker only need compromise that one location to obtain access to the usable file.

[004] To address reliability concerns, digital data is often "backed-up," i.e., an additional copy of the digital data is made and maintained in a separate physical location. For example, a backup tape of all network drives may be made by a small office and maintained at the home of a trusted employee. When a backup of digital data exists, the destruction of either the original device holding the digital data or the backup will not compromise the digital data. However, the existence of the backup exacerbates the security problem, as a malicious hacker can choose between two locations from which to obtain the digital data. Further, the site where the backup is stored may be far less secure than the original location of the digital data, such as in the case when an employee stores the tape in her home.

[005] Another method used to address reliability and performance concerns is the use of a Redundant Array of independent Drives ("RAID"). RAID refers to a collection of data storage schemes that divide and replicate data among multiple storage units. Different configurations of RAID provide increased performance, improved reliability, or both increased performance and improved reliability. In certain configurations of RAID, when digital data is stored, it is split into multiple units, referred to as "stripes," each of which is stored on a separate drive. Data striping is performed in an aigorithmically certain way so that the data can be reconstructed. While certain RAID configurations can improve reliability, RAID does nothing to address security concerns associated with digital data storage.

[006] One method that prior art solutions have addressed security concerns is through the use of encryption. Encrypted data is mathematically coded so that only users with access to a certain key can decrypt and use the data. Common forms of encryption include DES, AES, RSA, and others. While modern encryption methods are difficult to break, numerous instances of successful attacks are known, some of which have resulted in valuable data being compromised,

[007] Digitally stored data is subject to degradation over time, although such degradation tends to be extremely minor and the time periods involved tend to be much longer than for analog data storage. Nonetheless, if a single bit within a file comprised of millions of bits changes from a zero to a one or vice versa, the integrity of the file has been compromised, and its usability becomes suspect. Further, errors occur more frequently when digital data is transmitted due to noise in the transmission medium. Various prior art techniques have been devised to detect when a digital data segment has been compromised. One early form of error detection is known as parity, wherein a single bit is appended to each transmitted byte or word of data. The parity bit is set so that the total number of one bits in the transmitted byte or word is either even or odd. The receiving processor then checks the received byte or word for the appropriate parity, and, if it is incorrect, asks that the byte or word be resent.

[008] Another form of error detection is the use of a checksum. There are many different types of checksums including classic checksums, cryptographic hash functions, digital signatures, cyclic redundancy checks, and the use of human readable "check digits" by the postal service and libraries. All of these techniques involve performing a mathematical calculation over an entire data segment to arrive at a checksum, which is appended to the data segment. For stored data, the checksum for the data segment can be recalculated periodicaily, and checked against the previously calculated checksum appended to the data segment. For transmitted data, the checksum is calculated by the transmitter and appended to the data segment. The receiver then recalculates the checksum for the received data segment, and if it does not match the checksum appended to the data segment, requests that it be retransmitted.

[009] In 1979, two researchers independently developed a method for spiitting data among multiple recipients called "secret sharing." One of the characteristics of secret sharing is that a piece of data may be split among n recipients, but cannot be known unless at least t recipients share their data, where n > t. For example, a trivia! form of secret sharing can be implemented by assigning a single random byte to every recipient but one, who would receive the actual data byte after it had been bitwise exclusive orred with the random bytes. In other words, for a group of four recipients, three of the recipients would be given random bytes, and the fourth would be given a byte calculated by the following formula; s = s @r_a ®r_b ®r_c, where s is the original source data, r_a, r_b, and r_c are random bytes given to three of the four recipients, and s' is the encoded byte given to the fourth recipient. The original byte s can be recovered by bitwise exclusive-orring all four bytes together.

[0010] The problem of maintaining or reconstructing data stored on a digital medium that is subject to damage has aiso been addressed in the prior art. In particular, Reed-Solomon and Cauchy Reed-Solomon coding are two weil- known methods of dividing encoded information into multiple slices so that the original information can be reassembled even if all of the slices are not available, Reed-Solomon coding, Cauchy Reed-Soiomon coding, and other data coding techniques are described in "Erasure Codes for Storage Applications," by Dr. James S, Plank, which is hereby incorporated by reference.

[0011] Schemes for implementing dispersed data storage networks are also known in the art, In particular, United States Patent Number 5,485,474, issued to Michael O, Rabin, describes a system for splitting a segment of digital information into n data slices, which are stored in separate devices. When the data segment must be retrieved, only m of the original data slices are required to reconstruct the data segment, where n>m,

[0012] While dispersed data storage networks can theoretically be implemented to provide any desired level of reliability, practical considerations tend to make this impossible in prior art solutions. For example, dispersed data storage networks rely on storage media to store data slices. This storage media, like all storage media, will degrade over time. Furthermore, dispersed data storage networks rely on numerous transmissions to physically disparate slice servers, and data slices may become corrupted during transmissions. While TCP utilizes a CRC in every transmitted packet, the reliability provided by this CRC is not sufficient for critical data storage. Objects of the Invention

[0013] Accordingly, it is an object of this invention to provide a system, apparatus, and method for rebuilding data on a dispersed data storage network.

[0014] Another object of this invention is to provide a self-healing dispersed data storage network.

[0015] Other advantages of the disclosed invention will be dear to a person of ordinary skill in the art. It should be understood, however, that a system, method, or apparatus could practice the disclosed invention while not achieving all of the enumerated advantages, and that the protected invention is defined by the claims. Summary of the Invention

[0016] The disclosed invention achieves its objectives by providing an improved method for rebuilding a data segment stored on a dispersed data storage network when the data segment has been compromised in some manner. Generally, a dispersed data storage network maintains a data store of different data segments that are stored on the dispersed data storage network, where the term "data segments" means some quantity of stored data, and the term "data store" means any standard data storage mechanism, such as a file or database. When a data segment is stored, it is divided into some number of components, called data slices, and each data slice is stored on a separate slice server. Each data segment and each data slice is assigned a unique identifier. A data slice is compromised if it is, for example, outdated, corrupted, or missing, i.e., inherently not accessible by a slice server that is supposed to be storing it. A data segment becomes compromised when one or more data slices associated with the data segment become compromised. Note that in this context, the data segment becoming compromised does not mean that it cannot be read or rebuilt. It cannot be read or rebuilt only if too many data slices become compromised.

[0017] in one embodiment, the disclosed invention is a rebuilder application operating on a computer within a dispersed data storage network. The rebuilder application accesses a data store holding data segment identifiers for at least some of the data segments stored by the dispersed data storage network. The rebuilder application attempts to rebuild each data segment. For each data segment identifier, the rebuiider application identifies those slice servers that are supposed to store a data slice associated with the data segment identifier, The rebuilder application will also issue a request to each of the identified slice servers, which will respond with status data about the slice that the slice server is supposed to hold. The status data will indicate if the slice server can actually access the data slice it is supposed to store, or whether the data slice is "missing." In addition, the status data will include a transaction identifier indicative of the transaction on which the data slice was stored. Further, prior to sending the status data, each slice server may perform an integrity check on its stored data slice. If the integrity check fails, the status data will indicate that the data slice stored by the slice server is corrupted.

[0018] The rebuilder application examines the status data from the different slice servers that are supposed to hold data slices associated with the data segment identifier in question. If any of the status data indicates that a data slice is missing or corrupted, a record is added to a rebuild list identifying the compromised data segment and the compromised data slices. Further, all of the status data corresponding to a particular data segment are examined to determine the most recent transaction on which a data slice associated with the data segment was stored, and, if any of the other data sϋces were stored on an earlier transaction, a record is added to the rebuild list identifying the data segment and the outdated data slices.

[0019] The rebuilder application walks through the records on the rebuild list and rebuilds any compromised data siices. For each record on the rebuild list, the rebuilder application reads sufficient data slices from the dispersed data storage network to reconstruct the data segment identified in the rebuild record. It then slices the data segment according to an information dispersal algorithm, and writes any data slices that were identified as compromised in the rebuild record to the appropriate slice servers.

[0020] it should be understood that the transaction identifier discussed in the previous paragraph is used as an indication of the version of a stored data segment or slice. Accordingly, some other indication of version stored along with the data segment or slice would be equivalent to the transaction discussed above.

[0021] A second embodiment of the disclosed invention operates as a method of dealing with partially failed writes to a dispersed data storage network. In this embodiment, a data segment is sliced into m data slices, but only t of these data slices are successfully written during the initial write transaction, where r < m, and where no more than t data slices are required to reconstruct the data segment. In this circumstance, a rebuild record is created including the data segment identifier and the identifiers of any data slices that were not successfully written, and the data segment is then "rebuilt" as described above.

[0022] A third embodiment of the disclosed invention operates as a method of detecting compromised data slices during a read from a dispersed data storage network. In this embodiment, a list of slice servers is assembled, where each slice server is supposed to be able to access at least one data slice associated with the desired data segment. During the read, at least one of the slice servers from the list of slice servers returns status data indicating that the data slice it is supposed to store is compromised, in this circumstance, a rebuild record is created including the data segment identifier and the identifiers of any data slices that are indicated as compromised. The data segment is then rebuilt as described above.

Brief Description of the Drawings [0023] Although the characteristic features of this invention will be particularly pointed out in the claims, the invention itself, and the manner in which it may be made and used, may be better understood by referring to the following description taken in connection with the accompanying drawings forming a part hereof, wherein like reference numerals refer to like parts throughout the several views and in which: [0024] Figure 1 is a network diagram of a dispersed data storage network constructed in accordance with an embodiment of the disclosed invention; [0025] Figure 2 illustrates the use of checksums on a data segment as we!l as on an arbitrary number of data slices created from the data segment; [0026] Figure 3 is a flowchart illustrating the process by which a corrupted data segment can be rebuilt by a dispersed data storage network constructed in accordance with an embodiment of the disclosed invention; [0027] Figures 4A-4C collectively illustrate a read operation from a dispersed data storage network constructed in accordance with an embodiment of the disclosed invention; [0028] Figures 5A-5B collectively illustrate a write operation from a dispersed data storage network constructed in accordance with an embodiment of the disclosed invention; [0029] Figures 6A-6B collectively illustrate an automated process by which corrupted data slices may be recreated in accordance with an embodiment of the disclosed invention; [0030] Figure 7 is a flowchart illustrating the high-level operation of a rebuild process in accordance with an embodiment of the disclosed invention; [0031] Figure 8 is a flowchart illustrating a process for discovery of corrupt data occurring during the course of a normal read operation, in accordance with an embodiment of the disclosed invention; [0032] Figure 9 is a flowchart iilustrating a process for discovery of missing or outdated slices during the course of a normal read operation, in accordance with an embodiment of the disclosed invention; [0033] Figure 10 is a flowchart illustrating a process for discovery of missing slices during the course of a normal write operation, in accordance with an embodiment of the disclosed invention; [0034] Figure 11 is a flowchart illustrating a data scan that detects corrupt data slices, in accordance with an embodiment of the disclosed invention; [0035] Figure 12 is a flowchart illustrating a data scan that detects missing and outdated data slices, in accordance with an embodiment of the disclosed invention; [0036] Figure 13 is an example layout of a rebuild record, which stores the identities and other information used to manage the record of data slices that need to be rebuilt. I0037J Figure 14 is a flowchart illustrating a process that rebuilds missing, outdated, or corrupt data slices, in accordance with an embodiment of the disclosed invention; and

[0038] Figure 15 is an illustration of data that may be maintained by a rebuild process operating in accordance with an embodiment of the disclosed invention.

Detailed Description of the liiustrated Embodiment

[0039] Turning to the Figures, and to Figure 1 in particular, a distributed computer system implementing a dispersed data storage network 100 is shown. An arbitrary number of slice servers 150-162 store data slices sent to them by networked client computers 102,104,106. As illustrated, a grid access computer 120 allows access to the slice servers 150-162 by the client computers 102,104,106. In a dispersed data storage network, the grid access computer 120 acts as a router for data sent between a source computer and multiple slice servers 150-162. it manages transactions at the data segment level A manager appliance 122 monitors the activities and status of the grid access computer 120 and the slice servers 150 - 162. The manager appliance reports on statistics and events related to appliance health, network throughput, authentication attempts, read and write transaction counts, and rebuilding activities. It notifies users when events that could signal a problem occur,

[0040] As explained herein, the disclosed invention works to ensure the integrity of data stored in a dispersed data network not only by using checksums on each stored data segment as well as the constituent data slices, but also by reconstructing compromised data slices as well, in accordance with the disclosed invention, the grid access computer 120 will calculate a checksum for each data segment to be stored, and append the checksum to the data segment prior to slicing. The data segment is then sliced in accordance with an information dispersaf algorithm, and checksums are calculated and appended to each of the data slices. The data slices are then forwarded to siice servers 150-162, where the data slices are stored.

[0041] In addition, the access computer 120 also recreates data slices that have become corrupted, or were destroyed. If during operation of the dispersed data storage network 100, it is detected that a particular data slice has been corrupted or destroyed, a different data siice will be requested from a different slice server 150-162. Assuming that sufficient non-corrupted data slices exist to successfully reconstruct the original data segment, the reconstructed data segment will be re-sliced, and the corrupted data slice will be replaced with a non-corrupted version. Further, a rebuilder application operating within the dispersed data storage network periodically walks through all data slices stored on the dispersed data storage network. When a corrupted data slice is found, the rebuilder application identifies the data segment corresponding to the corrupted data siice, rebuilds the identified data segment, and rewrites the corrupted slice. Moreover, the rebuilder application actively engages in a detection process to identify corrupted, damaged, missing, and outdated data slices.

[0042] Figure 2 depicts the use of a checksum 220 on a data segment 230, as well as on the data slices 250-257 that the data segment 230 was divided into. Assuming that a data segment 230 is being written to a dispersed data storage network, a checksum 220 will be calculated for and appended to the data segment 230, thereby forming a "certified data segment." The certified data segment 230 will then be sliced as one piece of data resulting in data slices 250-257, i.e., when the data slices 250-257 are recombined, both the data segment 230 and data segment checksum 220 will be recovered. A checksum 240-247 is then calculated for, and appended to each data slice 250-257, forming "certified data slices" 260-267. The certified data slices 260-267 will then be sent to different siice servers. Figure 3 depicts one possible process by which corrupted slices may be recreated. During a read operation of the dispersed data storage network, a client requests a siice from a slice server in step 302. in step 303, the slice server transmits the requested slice to the client, which calculates a checksum for the requested data siice. In step 304, the calculated checksum is compared to the checksum appended to the stored data slice, and if the two checksums match, the read proceeds as normal in step 306. However, if the two checksums do not match, the slice server will transmit a message to a rebuilder application operating on the dispersed data storage network indicating that the requested data slice is corrupted in step 308, and return "Failure" to the querying server. In step 310, the grid access computer determines if an alternative slice can be read from a different slice server. If an alternative slice does not exist, the grid access computer will report an error in step 312. In step 314, the grid access computer reads the alternative slice. [0044] Figures 4A-4C show the process by which a dispersed data storage network, constructed in accordance with the disclosed invention and used in conjunction with the process depicted in Figure 3, coufd fulfill a read request, in step 402, a read request is received. Contained within the read request will be information sufficient to determine which slice servers contain applicable data, as well as the minimum number of data slices that must be retrieved before the requested data can be reconstructed. Further information on one method that can be used to associate data requests with particular slice servers is contained in United States Patent Application Number 11/973,621, entitled "VIRTUALIZED DATA STORAGE VAULTS ON A DISPERSED DATA STORAGE NETWORK," filed on October 9, 2007 and assigned to Cieversafe, inc. In step 404, the variable. m is initialized to the minimum number of slices required to construct the requested data segment. As described herein, for each successfully received and validated slice, m is decremented.

[0045] In step 406, a list of slice servers, each holding a required data slice that has yet to be received, is assembled. In step 408, the list is ordered by any applicable criteria. Further information on criteria by which the list may be ordered is contained in United States Patent Application Number 11/973,622, entitled "SMART ACCESS TO A DISPERSED DATA STORAGE NETWORK," filed on October 9, 2007 and assigned to Cieversafe, Inc. In step 410, read requests are issued to the first k slice servers on the assembled list, where k is at least equal to m, the minimum number of data slices needed to reconstruct the requested data segment, but could be as large as n, the number of data slices that have data relevant to the requested data segment. In step 412, r data slices are received, and in step 414 the number of received data slices r is subtracted from the variabie m. In step 416, m is compared to zero, and if m is greater than or equal to zero, execution returns to step 406 and proceeds as normal from there. However, if m is equal to zero, a collection of data transformations may optionally be applied to the received slices in step 418, The applied data transformations can include decryption, decompression, and integrity checking. In accordance with the disclosed invention, each data slice includes a cyclical redundancy check ("CRC"), or other form of checksum appended to the data contained in the slice. This checksum will be compared against a checksum calculated by the receiving slice server over the received data to ensure that the data was not corrupted during the transmission process. (n step 420, it is determined if the applied data transformations were successful for all of the received data slices. If the applied data transformations were not successful for some of the received slices, m is incremented by this number in step 422, and execution is resumed at step 406. The data transformations could fail, for example, if an integrity check revealed that a received data slice was corrupted. However, if the applied data transformations were successful for all received data slices, the received slices are assembled into the requested block of data in step 424. The same or different data transformations may optionally be applied to the assembled data block in step 426. Step 428 illustrates that the read process is completed. In accordance with the disclosed invention, a checksum for the data segment will be calculated and compared to a checksum appended to the assembled data segment.

[0047] In Figures 5A-5B the process by which a dispersed data storage network, constructed in accordance with the disclosed invention, could write data to a network of slice servers is depicted. In step 502 a data segment write request is received. Included in this write request is information sufficient to determine which slice servers the data segment should be written to, as well as information required by the information dispersal algorithm to divide the data segment, i.e., the number of slices to be written, referred to as n, as well as the minimum number of slices that are required to recover the data, referred to as m. Further information on one method that can be used to associate data writes with particular slice servers is contained in United States Patent Application Number 11/973,621 , titled "VfRTUALIZED DATA STORAGE VAULTS ON A DISPERSED DATA STORAGE NETWORK," filed on October 9, 2007 and assigned to Cleversafe, Inc.

[0048] A number of data transformations may optionally be applied to each block in step 506, and an information dispersal algorithm is applied in step 508. In particular, the Cauchy Reed-Solomon dispersal algorithm could be applied to the data segment, resulting in a predetermined number of data slices. In step 510, a number of data transformations are optionally applied to each data slice.

[0049] In the disclosed system, writes are performed transactionally, meaning that a minimum number of data slices t must be successfully written before a write is deemed complete. Normally, the number of data slices that must be successfully written wilf be set to the minimum number of slices needed to recreate the data. However, this number can be configured to a greater number, up to the number of slice servers in use. This would allow the user to continue using the dispersed data storage network during a minor network outage where one or more slice servers are unavailable. Slices that could not be immediately transmitted and stored couid be queued and transmitted when the network outage cleared, In addition, when a data segment is written to the dispersed data storage network, a transaction identifier is assigned and stored along with each written data slice. As explained later, this transaction identifier is used to ensure that the most recent version of a data segment has been stored to the dispersed data storage network, in step 512, a write transaction is initiated to the data storage network, As discussed herein, all slice servers are simultaneousiy contacted, and in step 514, a confirmation that at least t receiving slice servers are prepared to begin the write transaction, i.e., to store each slice, must be received, or the transaction is rolled back in step 516. In step 520 data slices are transmitted to the slice servers that indicated their ability to receive and store slices. The number of slice servers that successfully received and stored their assigned data slices is checked in step 522, and if less than t slices are successfully stored, the transaction is roiled back in step 516. If the result of step 522 is that the stores are successful, then a commit transaction is initiated in step 524 on all servers with successful writes. If the commit transaction fails, an error is logged in step 528. Otherwise, the write transaction was successful.

IS [0051] Figures 6A-6B are a flow chart illustrating an alternative process by which corrupted data slices may be recreated. In step 602, a scan of data slices is initiated by a rebuilder application operating somewhere on the dispersed data storage network, such as the grid access computer, the monitor appliance, or one or more slice servers. If no corrupted data slice is found, the corrupted slice recreation process is exited. However, if a corrupted slice is detected because of a failed integrity check, execution proceeds to step 806, where the rebuilder application determines what data segment corresponds to the corrupted data slice, and reads that data segment from the dispersed data storage network. The rebuilder application then reconstructs the data segment in step 608. In step 610, the data segment is re-sliced, and the rebuilder application rewrites a non-corrupted version of the corrupted data slice to the appropriate slice server in step 612. The process exits after the write has been completed. Detailed Description of Improved Rebuilder Application

[0052] The rebuilder application is responsible for ensuring that the integrity of all stored data segments is maintained. As such, the rebuilder application has access to a data store identifying every data segment stored by the dispersed data storage network. Note that referring to the rebuilder application as singular is solely for convenience; a system implementing the disclosed invention could be constructed using multiple rebuilder applications, each responsible for maintaining some subset of the stored data segments. [0053] Figure 7 illustrates the high-level operation of the rebuild process, and in particular, the interactions of the following components of the rebuilder application: the rebuild agent, the rebuild recorder, and the rebuild detector.

[0054] The rebuild agent executes rebuild operations. In order to rebuild a data segment the following operations are performed: 1) some or all of the available data slices for that data segment are read; 2) information dispersal algorithms are used to obtain a pre-dispersal form of the data segment; 3) information dispersal algorithms are used to generate restored versions of the previously missing/corrupted data slices; and 4) the restored data slices are written to the appropriate slice servers. When performing slice write operations, the rebuild agent will indicate the transaction identifier of the slices being written. The slice servers will use this identifier to ensure that slices are not overwritten if their transaction identifiers are greater than those specified.

[0055] The rebuild recorder stores information about data segments that have been identified as potentially needing to be rebuilt. This information is represented using "RebuiidRecords." A RebuildRecord consist of an identifier associated with the data segment to be rebuilt, the transaction identifier associated with the data segment to be rebuilt, and the identifiers of the data slices that store data associated with the data segment to be rebuilt. The rebuild recorder is responsible for providing rebuild records to rebuild agents, which actually perform the rebuilding operation. Figure 13 depicts one possible implementation of a RebuildRecord. Each RebuildRecord wili include one data segment record, which may include a data segment identifier, a transaction identifier, the status of any rebuild operation, a status description for any rebuild operation, the priority of any rebuild operation, and any other data associated with the data segment to be rebuilt. Each RebuildRecord will also include one or more data slice records, which may include a data slice identifier, the type of rebuild operation, a transaction identifier, the status of the rebuild operation, and a status description of the rebuild operation. The transaction identifier indicates the transaction on which the data segment or data slice was stored. Status indicates the progress of the rebuild operation; for example, PENDING, ACTIVE, COMPLETE, or FAILED. Rebuild type indicates how a particular data slice was compromised; for example, MISSING, OUTDATED, CORRUPTED, or UNKNOWN.

[0056] The rebuild detector actively discovers data slices that have been compromised in some way. For example, the rebuild detector is able to detect missing and outdated slices by downloading a list of slices from each siice server and comparing those lists. The rebuild detector can also detect corrupted data slices by verifying the checksums of alt data slices. This executes on each siice server in parallel,

[0057] In addition, the activities of the rebuiid detector, recorder, and rebuild agent generate statistics that are useful in monitoring the health of the dispersed data storage network. Examples of such statistics are number of RebuildRecords in the list, the time it takes to rebuild one slice, or the number of slices being rebuilt per second. These statistics can then be viewed on the manager appliance, or other similar monitoring agent.

[0058] in step 701, the rebuiid detector is triggered by some mechanism, such as the expiration of a timer based on configurable parameters related to frequency of rebuild, idle time in relation to other operations, and other parameters. The rebuild detector utilizes two separate types of scans.

[0059] In step 702, the rebuild detector scans by attempting to read and reconstruct each data segment from its constituent data slices. During the scanning process, the rebuild detector may notice that a particular data segment has data slices with different transaction identifiers, indicating that one or more of the data slices were not updated during a write, and therefore, that multiple versions of the same data segment are stored. The data slices with outdated transaction identifiers will be identified as compromised. It may also discover missing data slices. If it notes that a particular data slice in a data segment is missing or outdated, it passes the data slice to the rebuild recorder in step 705.

[0060] In step 703, the rebuild detector scans by looking directly at the slices on the slice servers, computing new checksums, and comparing to the stored checksum. If the computed checksum for a particular data slice does not match the checksum appended to the data slice, the identifying information for the data slice will be passed to the rebuild recorder in step 705,

[0061] In step 704, during normal read operations, if a missing, outdated, or corrupted data slice is read, the data slice identifier corresponding to the compromised data slice is passed to the rebuild recorder in step 705. In addition, during normal write operations, if a data segment cannot be written to all of the slice servers, the data slices that were not written are passed to the rebuild recorder in step 705. [0062] In step 705, the rebuild recorder generates the necessary data and forms or updates a RebuildRecord, which is appended to the rebuild list, based on the compromised data slices it has identified. In step 706, the rebuild recorder leases records from the list to a rebuild agent, which in step 707 rebuilds the data. The rebuilding of the data is done by reading enough slices to reconstitute a data segment, re-slicing the data segment, and storing the needed slices, resulting in a compiete and correct data segment,

[0063] Concerning the operation of the rebuild agent or agents, a single rebuild agent could handle all data slice rebuilding for a rebuilder application. Alternatively, a new process or thread could be created for each data slice to be rebuilt, in yet another alternative, a fixed stable of rebuild processes or threads could be spawned or instantiated when the rebuiider application was executed and rebuild records would then be passed to available rebuild agents as they finished rebuilding a compromised data slice.

[0064] Figure 8 depicts a normal read operation from a dispersed data storage network, illustrating in particular the process by which corrupted sitces may be discovered during the read. During the read operation, an access device requests a slice from a slice server in step 802. In step 803, the slice server transmits the requested slice to the access device, which calculates a checksum for the requested data slice. In step 804, the calculated checksum is compared to the checksum appended to the stored data slice, and if the two checksums match, the read proceeds as normal in step 806. However, if the two checksums do not match, the read operation will pass a message to the rebuilder application operating on the dispersed data storage network indicating that the requested data slice is corrupted in step 808. The rebuild recorder in step 809 inserts a RebuildRecord with the requested data slice's data slice identifier and an indication that the requested data slice is corrupted into its rebuild list so the identified data slice may rebuilt. In step 810, the grid access computer determines if an alternative slice can be read from a different slice server. If an alternative slice does not exist, the grid access computer will report an error in step 812. In step 814, the grid access computer reads the alternative slice Figure 9 depicts a normal read operation from a dispersed data storage network, illustrating in particular the process by which missing and outdated slices may be discovered. During the read operation, an access device requests a data slice from a slice server in step 901. In step 902, the data slice server returns the data slice or an error indicating that the requested data slice does not exist, if the requested data slice does not exist, in step 907 the rebuilder application is notified of the missing slice. Otherwise, in step 903, the slice server transmits the requested slice to the access device. In step 904, the transaction identifier of the data slice that was read is compared to the transaction identifiers of the other data slices that make up the data segment. If the transaction identifier matches the most recent transaction identifier of the other data slices used to reconstruct the data segment, the read proceeds as normal in step 905. However, if the transaction identifiers do not match, the read operation will pass a message to the rebuilder application operating on the dispersed data storage network indicating that the requested data slice is too old in step 908. The rebuild recorder in step 909 inserts or updates a RebuildRecord with the data slice Identifier corresponding to the outdated or missing data slice into its rebuild list so that the outdated or missing data slice may be rebuilt. In step 910, the grid access computer determines if an alternative data siice can be read from a different slice server. If an alternative data siice does not exist, the grid access computer will report an error in step 912. In step 914, the grid access computer reads the alternative data slice

[0066] Figure 10 depicts a normal write operation to a dispersed data storage network, illustrating in particular the process by which missing slices may be recorded during this process, in step 1001 , the normal write process begins, initiating concurrent transactional writes to the slice servers of all the data slices associated with a data segment. If the minimum required number of slice servers cannot be successfully written in step 1002, then in step 1003, the write transaction is rolled back, if, on the other hand, the minimum required number of slice servers are available, the data slices are written to the available slice servers. If the required number of data slices were written in step 1006, then a commit transaction is initiated in step 1010. If the commit succeeds on the required number of slice servers in step 1012 then the write is successful; otherwise, a write error is returned to the calling program in step 1014.

[0067] Step 1008 illustrates that if the write operation has determined that some number of slice servers were not able to write a data slice for a data segment, then the rebuild recorder is notified in step 1008 so that the missing data slices may be written in the future. In step 1009, the rebuild recorder inserts or updates a RebuildRecord for each missing data slice into its rebuild list so that the missing data slices can be "rebuilt" at a later time.

[0068] Figure 11 is a flow chart illustrating one of the scans performed by the rebuild detector, namely the scan for corrupted data slices. In step 1102, a scan of data slices held by each slice server is initiated by a rebuilder application operating somewhere on the dispersed data storage network. Once initiated, each slice server scans its own data slices by calculating a new checksum for each slice and comparing that checksum to the checksum appended to the stored data slice, if no corrupted data slice is found in step 1104, the corrupted slice detection process Is exited in step 1 105. However, if a corrupted siice is detected because of a failed integrity check, execution proceeds to step 1106, where the rebuild detector determines what data segment corresponds to the corrupted data slice, and reports that information to the rebuild recorder. In step 1108, the rebuild recorder inserts or updates a Rebuild Record including the corrupted data slice's information into its rebuild list.

[00691 Figure 12 is a flow chart illustrating the other scan performed by the rebuild detector, namely the scan for missing or outdated data slices. In step 1202, a scan of data slices is initiated by a rebuilder application operating somewhere on the dispersed data storage network. The rebuild detector asks each slice server for a list of the data slices contained on that server, in step 1204, the separate lists are processed in tandem with each other so that data slices from each data segment can be compared. If no missing or outdated data slice is found in step 1204, the missing or outdated slice detection process is exited in step 1205. However, if a slice is missing or a slice is determined to be older than the other slices in the data segment by comparing the transaction identifiers associated with the data slices, execution proceeds to step 1206, where the rebuϋder application reports information necessary to identify any outdated or missing data slices to the rebuild recorder. In step 1208, the rebuild recorder inserts or updates Rebuild Records corresponding to the missing or outdated data slices into its rebuild list so that the identified data siices may be rebuilt.

[0070] Figure 13 shows an example Rebuild Record, containing the data used to record data siices that need to be rebuilt.

[0071] Figure 14 illustrates the rebuilding of missing, outdated, or corrupt data slices which were detected and recorded in the preceding processes. The rebuild agent is responsible for this activity. In step 1401 a rebuild agent is initiated by a configurable time-based trigger, and requests data segments that need to be rebuilt from the rebuild recorder. In step 1402 the rebuild recorder responds with said data segments, if any exist. The data segments are prioritized, with the segments with the most compromised slices being sent first. For each data segment sent, in step 1403 the rebuild agent reads sufficient data slices from the slice servers to reconstruct the data segment. It then reconstructs the data segment in step 1404, and re-slices it in step 1406. Since the same information dispersal algorithm is used, the missing, outdated, or corrupt slices are recreated exactly as they were or would have been originally, In step 1408, the missing, outdated, or corrupt data slices are written to the appropriate slice servers. When this is complete, the rebuild agent returns the results to the rebuild recorder so the rebuild recorder in step 1410 can make its list up to date. It deletes rebuild records that were successfully rebuilt. It un-!eases the records which were not successfully rebuilt so that they will be provided again in the next iteration of the rebuild agent process. Figure 15 illustrates the specific operation of the rebuild detector. As depicted, data concerning five data segments has been gathered by the rebuild detector, which actively queries slice servers about data segments which the slice servers store. Table 1 shows the transaction identifier associated with data slices held by three slice servers for five separate data segments. Further, for the depicted system, at least two data slices are required to rebuild a data segment. Rows 1512,1514,1516,1518,1520 correspond to the stored data segments, and columns 1502,1504,1506 correspond to the slice servers holding the data slices. For data segment 0, which is depicted by row 1512, the data slices stored by all three slice servers were stored during transaction 100, and therefore, the transaction identifier associated with each data slice is 100, As none of these data slices are outdated or missing, none of these data slices will be noted as compromised. However, in regards to data segment 1 , slice servers A and C hold data slices that were written during transaction 99, while slice server B holds no data slice at all. This may have occurred because an extended outage prevented the data slice from being written at all, or the portion of a hard disk holding the data slice may have failed. Accordingly, the data slice that should have been held by slice server B is noted as compromised and added to the rebuild list. It may further be marked as missing as opposed to outdated or corrupted.

[0073] Data segment 2, which is represented by row 1516, aiso has no outdated, missing, or corrupted data slices, and therefore, no entry wili be made in the rebuild list corresponding to data segment 2. However, in regards to data segment 3, which is represented by row 1518, the data slice stored by slice server C was stored during transaction 100, while the data slices stored by slice servers A and B were stored during transaction 101. Accordingly, the data slice stored by slice server C is likely outdated, and is added to the rebuild list.

[0074] Data segment 4 illustrates a case where a stored data segment cannot necessarily be rebuilt, in this case, the data slice stored by slice server A was stored during transaction 102, while the data slice stored by slice server B was stored during transaction 99. In addition, the data slice held by slice server C has become corrupted. As a minimum of two data slices are required to reconstruct a data segment in this example, and only one fully updated data slice is available, it is possible that data segment 4 may no longer be rebuildable. Nonetheless, as two data slices are available, albeit one of them may be outdated, a rebuild operation will be attempted. As with all rebuilt data segments, after a data segment is reconstructed using the correct information dispersal algorithm, the checksum of the rebuilt data segment is computed and checked against the checksum appended to the data segment. Assuming the checksums match, the data segment is intact, and it wili be resϋced, and the transaction number for each data slice set to the most recent transaction, i.e., 102 in this case.

10075] As illustrated, the process of detecting missing and outdated data slices involves comparing the data slices stored by each slice server. As the number of stored data segments may be extremely large, a complete representation of every stored data slice may be too large to hold in memory. Thus multiple iterations, each producing a partial list of stored data slices, may be required in order to process all stored slices. Such a process would proceed on a data segment by data segment basis, with information about all data slices comprising some number of particular data segments being pulled and analyzed during each iteration.

[0076] The foregoing description of the invention has been presented for purposes of illustration and description, and is not intended to be exhaustive or to limit the invention to the precise form disclosed. The description was selected to best explain the principles of the invention and practical application of these principles to enable others skilled in the art to best utilize the invention in various embodiments and various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention not be limited by the specification, but be defined by the claims set forth below.

Claims

ClaimsWhat is claimed is:

[C1] 1. A method of rebuilding data on a dispersed data storage network comprising a plurality of slice servers and at least one grid access computer, said method comprising the steps of: i) accessing a data store holding a list of data segment identifiers, said data segment identifiers corresponding to data segments stored by said dispersed data storage network; and ii) for at least one data segment identifiers in said list of data segment identifiers:

(1) identifying a list of slice servers potentially storing a data slice associated with said data segment identifier;

(2) retrieving status data from each of said slice servers, said status data corresponding to a data slice potentially held by said slice server;

(3) noting that at least one data slice holding data associated with a data segment has been compromised based on said status data,

[C2] 2. The method of claim 1 wherein said status data includes an indication that the corresponding data slice is corrupted and wherein said step of noting is based on said indication.

[C3] 3. The method of ciaim 1 wherein said status data includes an indication that the corresponding data siice is not stored by the slice server from which the status data was retrieved and wherein said step of noting is based on said indication.

[C4] 4. The method of claim 1 wherein said status data includes a transaction identifier indicating the transaction on which the corresponding data slice was stored.

[C5] 5. The method of claim 4, further comprising the step of for said at least one data segment identifier, determining the transaction identifier corresponding to the most recent transaction on which a data slice corresponding to said at least one data segment identifier was stored, and wherein said step of noting is based on determining that at least one data slice was stored prior to said most recent transaction.

[C6] 6. The method of claim 1 further comprising the steps of; i) determining at least one data segment identifier where at least one data slice holding data associated with said data segment identifier has been compromised; ii) adding said at least one data segment identifier to a rebuild list; iii) reading sufficient data slices from said dispersed data storage network to reconstruct a data segment associated with said data segment identifier; iv) slicing said data segment according to an information dispersal algorithm; and v) writing at least one new data slice corresponding to said at least one noted as compromised data slice.

[C7] 7. A method of writing data to a dispersed data storage network including a plurality of slice servers, said method comprising the steps of: i) receiving a data segment to be written to said dispersed data storage network;

U) slicing said data segment into a quantity m of data slices; iii) simultaneously transmitting said m data slices to m slice servers so that each data slice is transmitted to a separate slice server; iv) storing t data slices so that each stored data slice is accessible by a separate slice server, and where t < m; and v) noting that at least one data slice holding data associated with said data segment was not written to a slice server.

[C8] 8, A method of reading data from a dispersed data storage network including a plurality of slice servers, said method comprising the steps of; i) receiving a read request for a data segment stored on said dispersed data storage network; ii) assembling a list of said slice servers, wherein each slice server on said list can potentially access at least one data slice associated with said data segment; iii) receiving from at least one of said slice servers status data indicating that said at least one data slice potentially stored by said at least one slice server is compromised; and iv) noting that at least one data slice associated with said data segment has been compromised. [C9] 9. The method of claim 8 wherein said status data includes an indication that the corresponding data slice is corrupted and wherein said step of noting is based on said indication. [C10] 10. The method of claim 8 wherein said status data includes an indication that the corresponding data slice is not stored by the slice server from which the status data was retrieved and wherein said step of noting Is based on said indication. [C11] 11. The method of claim 8 wherein said status data includes a transaction identifier indicating the transaction on which the corresponding data slice was stored,