WO1999009480A1

WO1999009480A1 - Method and system for nonredundant backup of identical files stored on remote computers

Info

Publication number: WO1999009480A1
Application number: PCT/IB1998/001203
Authority: WO
Inventors: Cary Swoveland; Robert Somerville
Original assignee: Telebackup Systems, Inc.
Priority date: 1997-07-29
Filing date: 1998-07-22
Publication date: 1999-02-25
Also published as: AU8457698A

Abstract

Method for backing-up files stored on remote computers to a central computer. The method includes storing a file list in a memory device accessible to the central computer. The file list contains multiple records which each correspond to a file stored at some time on one or more of the remote computers. Each record includes file identification data which identifies a respective file and includes a signature of the respective file. During a backup operation of a remote computer, file identification data is transmitted from the remote computer to the central computer. The central computer compares the file identification data, including the file signature, received from the remote computer with the file identification data of one or more records contained in the file list. If the file identification data received from the remote computer does not match the file identification data of any of the records contained in the file list, the central computer transmits a message to the remote computer instructing the remote computer to transmit the file to the central computer. The central computer further adds a record containing the file identification data to the file list.

Description

METHOD AND SYSTEM FOR NONREDUNDANT BACKUP OF IDENTICAL FILES STORED ON REMOTE COMPUTERS

COPYRIGHT NOTICE

A portion of the disclosure of this patent document contains material which is

subject to copyright protection. The copyright owner has no objection to the facsimile

reproduction by anyone of the patent document or the patent disclosure, as it appears in

the Patent and Trademark Office patent files or records, but otherwise reserves all

copyright rights whatsoever.

BACKGROUND OF THE INVENTION

The invention disclosed herein relates generally to computer systems for backing

up files. More particularly, the present invention relates to a method and system for

backing up files stored on a large number of remote computers to a central computer

while reducing storage and bandwidth requirements by minimizing or eliminating the

redundant storage of identical files stored on more than one of the remote computers.

Data stored in the hard drive files of personal computers and workstations are

generally inadequately protected against loss. The danger of loss can arise when the hard

drive becomes faulty through failure, destruction by fire, flood or other physical disasters

as well as the data becoming corrupted, lost or, infected by a computer virus. Typical

backup methods for hard drives such as floppy disk media, attached tape drives, or

duplicate hard drives require substantial manual initiation to set-up, define and operate, storage media handling including off-site transportation for storage and retrieval, and

dedication of the workstation users prime time to oversee the whole backup process.

Remote computer backup can be used to move copies of data files from one

computer over a transmission medium such as a telephone line to an offsite computer and

storage medium. A remote backup guards against many of the problems which give rise

to loss of data in a computer's hard drive. The protection is provided by the presence of a

number of previous copies of the files which can be restored to the workstation based on

the date of the desired copy.

Attempts to unburden the workstation user through automatic on-line processes

have consistently run into communications issues, the prime one being the amount of

stable connected time (bandwidth) to achieve the initial complete backup and subsequent

incremental backups on a regular cycle. Additionally, the amount of storage required has

had a significant impact on media selection, number of backup sets maintained, and cost

of maintenance.

Existing remote backup systems have not been able to effectively reduce the

communications time required for the first full backup of a new workstation and

subsequent backups, and the amount of storage required to perform the backup. There is

thus a need for a backup process which is designed to solve the problems associated with

backups of remote computers and to reduce the bandwidth and storage requirements

associated with existing backup processes. BRIEF SUMMARY OF THE INVENTION

It is an object of the present invention to provide an improved backup method and

system which solves the problems set forth above associated with existing backup

systems.

It is another object of the present invention to reduce the bandwidth required for

the backup of files stored on a remote computer to a central computer.

It is another object of the present invention to reduce the storage requirements in a

central computer backing up a number of remote computers.

It is another object of the present invention to minimize or eliminate the redundant

storage on ?. central computer of identical files backed up from a number of remote

computers.

The above and other objects are achieved by a method for backing up files

stored on remote computers to a central computer, the remote computers being

connectable to the central computer. The method involves storing a file list in a memory

device accessible to the central computer, the file list containing a plurality of records

each record corresponding to a file stored at some time on one or more of the remote

computers, and each record comprising file identification data identifying a respective file

and including a signature of the respective file. During a backup operation of a first

remote computer, first file identification data is transmitted from the first remote

computer and received at the central computer. The first file identification data identifies a first file stored at some time on the remote computer and includes a signature of that

file.

The central computer compares the first file identification data including the first

file signature received from the first remote computer with the file identification data of

one or more records contained in the file list. If the first file identification data received

from the first remote computer does not match the file identification data of any of the

records contained in the file list, the central computer transmits a message to the first

remote computer instructing the first remote computer to transmit the first file to the

central computer. The central computer further adds a record containing the first file

identification data to the file list.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention is illustrated in the figures of the accompanying drawings which are

meant to be exemplary and not limiting, in which like references refer to like or

corresponding parts, and in which:

Fig. 1 is a block diagram showing the backup system of one embodiment of the

present invention;

Figs. 2A-2C show the sequence of messages transmitted between the central

computer and remote computer during a backup operation in accordance with one

embodiment of the present invention; Figs. 3A-6 contain flow charts showing the method of backing up a remote

computer onto the central computer of the system shown in Fig. 1 in accordance with one

embodiment of the present invention, of which:

Figs. 3A-3B show the process performed by the remote computer of

creating changed image records for transmission to the central computer;

Figs. 4A-4B show the process performed by the central computer of

processing the changed image records received from the remote computer as a result of

the process shown in Figs 3A-3B and creating modified changed image records for

transmission to the remote computer;

Figs. 5A-5B show the process performed by the remote computer of

processing the modified changed image records received from the central computer as a

result of the process shown in Figs. 4A-4B and selecting files to be transmitted to the

central computer for backing up; and

Fig. 6 shows the process performed by the central computer of storing the

files received from the remote computer as a result of the process shown in Figs. 5A-5B;

Figs. 7A-7C contain a flow chart showing the process of restoring files to the

remote computer; and

Figs. 8-15 are screen displays displayed by one version of a software program

implementing an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The method and system of the present invention will be described herein with

reference to the system diagrams shown in Figs. 1 and 2, the flow charts shown in Figs.

3A-7C, and the screen displays shown in Figs. 8-15. One skilled in the art will recognize

that these drawings illustrate one manner of implementing the present invention and that

many variations are possible within the scope of the present invention.

As shown in Fig. 1, the system of one embodiment contains a central computer 10

and a number of remote computers 12 connected to the central computer 10. The central

computer 10 may be a single computer or a group of networked computers, operated as

the backup file repository. The remote computers 12 may be personal computers,

workstations attached over a network, network servers, or other types of computers or

computer systems. The remote computers 12 may be connected to the central computer

by virtue of any conventional communications link, including copper wire network,

coaxial cable network, satellite network, fiber optics network, or microwave network.

Information may also be transmitted between the central and remote computers by

physical media transfer such as by disk, tape or other information storage device. As

such, the remote computers 12 are not necessarily distanced from the central computer 10

by any specific amount.

The remote computers 12 (a representative one being shown schematically in Fig.

1) contain storage devices such as direct access storage devices on which are stored a

number of files 14. In accordance with well-known practice, each file 14 has associated therewith a file descriptor consisting of a record containing information about the file

(other than the contents of the file itself) that is stored by the disk operating system of the

remote computer 12. File descriptors includes the name of the file, the location of the file

in the file system (such as the drive and directory), the size of the file, and its date/time

stamp. The size of the file is usually represented by the number of bytes in the file. The

date/time stamp usually represents the date and time the file was last changed, regardless

of whether the file was last changed while on the remote computer 12 or last changed on

another computer and copied to the remote computer 12. If the file was never changed

(after it was created) on the remote computer 12 or some other computer, the date/time

stamp is the date and time the file was created. Depending upon the operating system, the

file descriptor may also include the date and time the file was created, the type of file

(e.g., read-only, system, hidden), whether an archive bit is set or not, and extended file

attributes as known to those of ordinary skill in the art.

For purposes of the system of the preferred embodiments, two files 14 are

considered to be different when one file does not contain exactly the same information as

the other file. Because the focus is on the content of the files, two files with different file

names or other identifying characteristics may not differ in this sense. To distinguish

between files, the system computes a signature for each file 14. In preferred

embodiments, a file's signature is the numeric value of a mathematical function that is

applied to all the bytes in the file. This function belongs to a class of polynomial

functions called "cyclic redundancy checks," which are well-known functions developed to detect errors in modem transmissions. The particular polynomial function used in one

preferred embodiment produces two integers ~ one four byte, one two byte - which

combined represent any value between zero and approximately one trillion. As one

skilled in the art will recognize, however, other mathematical functions could be used for

the purpose of computing a file's signature, e.g., a file digest, exclusive-oring, etc. The

particular function used for the signature represents a balance between accuracy and

speed as understood by those of skill in the art. In accordance with the invention, two

files are considered different by virtue of their signatures being different and two files are

considered identical by virtue of at least having identical signatures and, in preferred

embodiments, by virtue of having identical sizes and signatures.

The central computer 10 has storage devices 16 for storing backup copies of files

14 stored on the remote computer 12. The storage devices 16 may be any conventional

storage devices including magnetic storage such as hard disks and optical storage such as

writable CD-ROMs. In some embodiments, the storage devices 16 consist of two or more

different types of devices having differing capabilities to serve as primary and secondary

storage devices. For example, the storage devices 16 of one embodiment are high

capacity (4.25 gigabyte), SCSI high speed on-line hard drive storage disks serving as

primary storage and high capacity (8 gigabyte), 4 mm DAT tape drives for disaster

recovery. The capacity and number of such storage devices depends upon the needs of

the system as would be understood by those of skill in the art. In accordance with one aspect of the present invention, the central computer 10

serves as a "common file archive" for all the files 14 stored on all remote computers 12

connected or connectable to the central computer 10. That is, the central computer

maintains a virtual repository of files which are the same for two or more remote

computers. These files are not physically transferred from the second and subsequent

remote computers which contain them. This common file archive ensures that the only

files physically transferred from a remote computer are truly unique to that computer.

Hence, bandwidth capacity to accommodate the backup process is minimized as well as

the storage required for the backup set.

To implement this common file archive, the central computer 10 contains a full file

list 18 ("FFL"), a file ownership list ("FOL") 20, and a common file backup processing

program 22. The FFL 18 is a set of records 24 each corresponding to a file 14 that is or

was located on one or more remote computers 12. Each FFL record 24 contains the file

descriptor for the corresponding file 14, a signature of the file 14, a code identifying the

remote computer 12 operating system from which the file 14 was backed up, and a FFL

File ID that is unique to each FFL record which represents a file stored on one or more of

the remote computers 12.

The FOL 20 is a set of records 26 each of which corresponds to a particular remote

computer and either a file that is or was located on that computer, or a record of a file on

the remote computer that has been erased or overwritten. Each record contains a unique

identifier for the remote computer, which is assigned by the central computer for each remote computer to which it is or may be connected, the FFL File ID for the associated

record in the FFL, and the date and time the file was backed up on the remote computer,

or was recognized as having been newly-erased, as explained further below.

In accordance with the invention, and as explained in greater detail below, the FFL

18 is used by the central computer 10 to determine whether a file 14 to be backed up by a

given remote computer 12 is already backed up in the backup storage 16 from another

remote computer 12, and the FOL 20 is used to keep track of which of the files 14

identified by the FFL records 24 are stored on each remote computer 12 at a given time.

Both the FFL 18 and FOL 20 may be stored in the central computer 10 as a set of records

or as entries in a one or more databases or tables.

The common file backup processing program 22 manages the backup processes

performed on the central computer 10, as explained below. The central computer also has

a restoration program or routine 28 which manages the process of restoring a given

remote computer 12 following a hard drive or other failure or data loss.

In preferred embodiments, each remote computer 12 has a full image ("FI") list 30

and an incremental backup program 32. The FI 30 is a set of records 34 maintained on

each remote computer 12 that stores information regarding files which have previously

been backed up on the central computer 10 from that remote computer 12. The FI

contains all information contained in the records 24, 26 in the FFL 18 and FOL 20,

respectively, relating to that particular remote computer 12, except for the FFL File ID.

The FI 30 is used by the incremental backup program 32 during a backup operation performed at the remote computer 12, as described further below, to determine whether

each file 14 then stored on the remote computer 12 has already been backed up on the

central computer 10, and whether any files previously backed up to the central computer

10 from this remote computer 12 have since been erased from the remote computer 12.

The product of the backup program 32 is a set of changed image ("CI") records 36.

The CI records 36 represent the changes to the files 14 stored on the remote computer 12

since a prior backup operation to the central computer 10. Each CI record contains the

file descriptor and file signature for each file whose status has changed since the prior

backup, and a status field indicating whether the specified file is new, changed, or newly-

erased. These file status options, as used in the preferred embodiments, have the

following meanings:

(i) a file 14 is considered "new" to a given remote computer 12 if it was not

present in the same location in the remote computer's file system when the remote

computer 12 was last backed up. For most computer operating systems, this means that

the file 14 was not present in the same directory on the same logical drive when the

remote computer 12 was last backed up;

(ii) a file 14 is considered "changed" to a given remote computer 12 if it differs

from, but has the same name, as a file that was present in the same location in the remote

computer's file system when the remote computer 12 was last backed up. For most

remote computer operating systems, this means that there was a different file by the same name in the same directory on the same logical drive when the remote computer was last

backed up; and

(iii) a file 14 is considered "newly-erased" to a given remote computer 12 if it was

present when the remote computer 12 was last backed up, but is no longer present in the

same location in the remote computer's file system. For most remote computer operating

systems, this means that the file that was present when the computer was last backed up is

no longer present in the same directory on the same logical drive.

In preferred embodiments, a single byte is used for the file status field, using the

representations "N" for "new," "C" for "changed," and "E" for "erased."

With respect to these status types, a file 14 on the remote computer 12 is

considered to be "backed up" when an exact copy of the file 14 and the file descriptor is

stored on the central computer 10, and a record of the existence of the file 14 and file

descriptor on the remote computer 12 has been created and stored on the central computer

10.

An overview of the sequence of communication between the central computer 10

and a remote computer 12 during a backup operation is shown in Figs. 2A-2C. Each step

in the process is described in more detail below. As shown in Fig. 2A, the remote

computer 12 produces a set of CI records 36 and transmits these records to the central

computer 10.

The central computer 10 processes the CI records 36 and produces modified CI

records 36P and 36L which are transmitted to the remote computer 12 as shown in Fig. 2B. Some modified records 36P contain a field with an instruction to the remote

computer 12 to perform a physical transfer of the file 14 identified by the CI record 36. A

physical backup is the process of transmitting a copy of a file 14, together with its file

descriptor, from the remote computer 12 to the central computer 10, where the file is

stored, and creating and storing on the central computer 10 a record of the existence of

this file 14 and its file descriptor. Other modified CI records 36L contain a field with an

instruction to the remote computer 12 to perform a logical transfer of the file 14 identified

in the CI record. A logical backup is the operation of transmitting a copy of a file

descriptor from the remote computer 12 to the central computer 10 (but not the file itself),

where the file descriptor is stored, and creating and storing a record of the existence of the

file descriptor and associated file on the remote computer 12.

As shown in Fig. 2C, the remote computer 12 responds to the modified CI records

by transmitting the files 14 identified in the modified CI records 36P, which contained an

instruction to perform a physical transfer, to the central computer 10. To reduce

bandwidth and provide for security and privacy, the files 14 are compressed and

encrypted before transmission.

Each step in this backup process is described in more detail with reference to the

flow charts in Figs. 3A-6.

The process of creating the CI records is shown in Figs. 3A-3B. The backup

process is started at the remote computer, step 50. The process can be started in one of

two ways: (i) the operator of the remote computer 12 can cause the backup program to be

executed, in the same way the execution of other application programs are initiated; or

(ii) a program running in the background on the remote computer 12 can initiate

the backup operation at a particular time-of-day which may be scheduled and modified by

the operator of the remote computer 12.

Each file 14 stored on the remote computer 12 is then retrieved, step 52, and a

signature is computed for the file, step 54. If the remote computer 12 has not previously

been backed up to the central computer, all the files 14 would be new, and a CI record is

created for each file containing the respective file's file descriptor, file signature and a

status field set to "new."

If a prior backup operation has been performed for the remote computer 12, the

file descriptor for each file 14 is compared to the file descriptors in the FI records, step

56. In particular, the file name and file location parts of the file descriptor are used in this

comparison. If no file name and location in the FI records' file descriptors match a given

file's file name and location, step 58, the file status is set to "new," step 60, and a CI

record is created with the file's file descriptor, file signature, and file status of "new," step

62 (Fig. 3B).

If the file name and location for a given file 14 match a file name and location of a

FI record, step 58, then it is understood that the file 14 or a version thereof has previously

been backed up to the central computer 10. A flag associated with the FI record is set,

step 64, indicating that a file corresponding to this FI record exists and has not been erased from the remote computer 12. Tests are then performed to determine whether the

file 14 has changed since the last backup. In some embodiments, this determination is

made through use of the size and signature of the file. That is, the size of the file is

compared to the file size stored in the FI record's file descriptor, step 66. If the file sizes

do not match, then the file 14 has changed since the last backup. If the file sizes match,

the file signature of the file is compared to the file signature stored in the FI record, step

70. If the file signatures are different, then the file has changed. In either case of non-

matching values, the file status is set as "changed," step 68, and a CI record is created

with the file descriptor, file signature, and file status of "changed," step 62.

If the file signatures match, then the file is considered not to have changed since

the previous backup operation. As a result, no CI record is created for this file. If there

are any more files, step 72, the next file is processed in the same manner.

In alternative embodiments, only the file sizes are compared, thus saving time by

avoiding the need to compute file signatures for all files stored on the remote computer.

In further embodiments, file signatures are only computed and compared for those files

having a file size which matches a file size stored in a FI record. The remote computer

user may be given the option of using file signatures to achieve greater accuracy in the

comparison process.

When all remote computer files 14 have been so processed, the FI records are

checked, step 74, to determine whether any FI records were not accessed, thus showing

that the files represented by those FI records have been erased. For each FI record, the FI record accessed flag is checked, step 76. If the flag is not set, then the file status is set to

"erased," step 78, and a CI record is created containing the file descriptor and file

signature from the FI record and a file status of "erased," step 80. If there are any more

FI records to check, step 82, the same process is performed on each.

When this process is complete, a set of CI records will have been created

representing files which are new, changed and erased since the prior backup, if there had

been one. The date and time at which the backup operation was initiated on the remote

computer is attached to each CI record. The CI records 36 are transmitted to the central

computer 10, step 84.

One skilled in the art will recognize that the process just described of identifying

files which are new, changed or erased may be performed entirely on the central

computer 10 by transmitting the file descriptors and file signatures for all files 14 stored

on the remote computer 12 to the central computer 10. However, performing this process

on the remote computer 12 reduces the amount of data needed to be transmitted to the

central computer 10, thus reducing the bandwidth required in the system. Furthermore,

other means could be used to determine whether a file on the remote computer 12 has

changed, such as by examination of the file's "archive bit," as understood by those of

skill in the art, if this information is included in the file's file descriptor.

The central computer 10 processes the CI records 36 in accordance with the

procedure shown in Figs. 4A-4B. The central computer 10 receives the CI records, step

100, and determines the remote computer ID, step 102. The remote computer ID may be found in a look up table or similar data structure stored on the central computer 10, or

may be attached as one of the first parameters sent by each remote computer 12 with a set

of CI records. The central computer opens the FFL and FOL, step 104, which will be

used to determine whether any of the files identified by the CI records were already

obtained from another remote computer 12 and are already stored in the backup storage

16.

Each CI record is opened, step 106, and the status field checked, step 108. If the

file status is set to "E" for "erased" (meaning that the file identified by the CI record is

stored in the backup storage), the FFL is checked to find the FFL record which

corresponds to the CI record. In one embodiment, this is accomplished by retrieving each

FFL record, step 110, and comparing the FFL record's file descriptor to the file descriptor

in the CI record, step 112. If the file descriptors match, the file signatures from the FFL

record and CI record are compared, step 114. If the file signatures match, then the current

FFL record represents the file which has been erased on the remote computer 12, and

another record is added to the FFL which is a duplicate of the current FFL record, step

116. The duplicate record is inserted into or indexed in the FFL in a position immediately

following the current FFL record. This duplicate, "deletion" record is used during the

restoration process in a manner described below. An action code, which will be stored in

the FOL, is set to "newly erased," step 118.

In preferred embodiments, the original FFL record is not immediately deleted from

the FFL, nor is the compressed file stored in backup storage 16 immediately deleted. Rather, a counter is associated with each FFL record which counts the number of remote

computers storing the file identified by the FFL record. The counter is incremented

(increased by one) each time a backup operation is performed as described herein in

which a remote computer transmits a CI record indicating that it is storing a "new" file

corresponding to the FFL record. The counter is decremented (decreased by one) each

time a remote computer sends a CI record indicating that the corresponding file has been

erased or changed from a remote computer. If a FFL record's counter is non-zero, the

FFL record and corresponding compressed file are retained by the central computer.

When a counter reaches zero, the FFL record and corresponding compressed file stored in

backup storage are either deleted immediately, at a scheduled time set by a system

administrator with other zero-counter FFL records, or when the FFL record has reached a

certain age as set by the system administrator and determined by reference to the

corresponding FOL record.

If the file status field is not equal to "E" for "erased," step 108, then the CI record

represents a new or changed file. The FFL is then checked to determine whether the

identical file is already stored in the backup storage 16. In particular, each FFL record is

retrieved, step 120, and the file size and signature from the FFL record is compared to the

file size and signature from the CI record, step 122. If the file sizes and signatures match,

the file descriptors from the FFL record and CI record are compared, step 124. If the file

descriptors match, the file identified by the CI record is deemed to be identical to a file

already stored in the backup storage 16. Thus, a copy of the file 14 need not be transmitted from the remote computer 12 to the central computer 10. To indicate this, a

"logical transfer" code is added to the CI record, step 126, thus creating a modified CI

record which will be transmitted back to the remote computer 12. In addition, the action

code, used in the FOL, is set to "file backed up," step 128.

If either the file size/signature or file descriptors from the CI record do not match

any FFL record, the file 14 is deemed to be new, i.e., not already stored in the backup

storage 16. A "physical transfer" code is added to the CI record, step 130. to serve as an

instruction to the remote computer 12 to physically transfer the file to the central

computer 10 for backup.

One skilled in the art will recognize that any suitable search routine may be used

for locating FFL records having values that match those in a given CI record, including

sorting, queries, binary tree searches, etc. Furthermore, one skilled in the art will

recognize that different parameters from the CI records and FFL records may be

compared, or may be compared in a different sequence, provided that the file signature

from a CI record is compared to the file signature in the FFL record before a

determination is made that the files are identical.

For erased and changed files, a record is added to the FOL, step 132, containing

the remote computer ID, FFL File ID of the current FFL record (which matched the CI

record), and action code and the current date and time. This FOL record is used to

identify the state of files on the remote computer as of a given date and time, and is used

during the restoration process. In some embodiments, a FFL and FOL record is also created at this point for new files, but a flag is set for each record to indicate that the file

has not yet been received. When the file is received from the remote computer, the flag is

reset to zero. In other embodiments, the FFL and FOL records for new files are not

created until the file is actually received from the remote computer, as shown in Fig. 6

and discussed below.

If all CI records have been similarly processed, step 134, the modified CI records

are transmitted from the central computer 10 to the remote computer 12, step 136. The

modified records instruct the remote computer 12 what action to take with respect to each

file identified by a CI record.

The process performed on the remote computer 12 is shown in Figs. 5A-5B. The

remote computer 12 receives the modified CI records, step 150, and opens each, step 152.

If the modified CI record contains a record code of "physical transfer," step 154, the

remote computer 12 retrieves the file 14 having a file descriptor equal to the file

descriptor in the modified CI record, step 156. The remote computer 12 compresses the

file, step 158, and encrypts it if desired. The compressed file 14 is transmitted to the

central computer 10 along with the file descriptor and signature, step 160. An action code

to be placed in a new FI record is set to "backed up," step 164.

If the modified CI record contains a "logical transfer" code, step 162, then the

remote computer 12 does not send the file to the central computer, but simply sets the

action code to "backed up," step 164. If the modified CI record contains neither a "physical transfer" or "logical transfer" code, then the file identified by the CI record was

erased, and the action code is set to "erased," step 166.

For each modified CI record, a FI record is added, step 170, containing the file

descriptor and file signature from the modified CI record, the action code, and the date

and time the backup was initiated, which was attached to the CI record as explained

above. New FI records with action codes of "erased" are preferably stored or indexed to

immediately follow the corresponding FI record with the same file descriptor and

signature. If there are no other modified CI records, step 172 (Fig. 5B), a message is

transmitted to the central computer 10 that "backup is complete," step 174.

The central computer processes the compressed files in accordance with the

process shown in Fig. 6. For each file, the central computer receives the file descriptor,

file signature and compressed (and possibly encrypted) file 14, step 190. The compressed

file 14 is stored in backup storage 16, step 192, and new FFL and FOL records are created

by assigning a new FFL File ID, step 194, adding a record to the FFL containing the file

descriptor, file signature and FFL File ID, step 196, and adding a record to the FOL

containing the remote computer ID, FFL File ID, an action code of "File backed up," and

the date/time the backup operation was initiated (as attached to the CI record as explained

above), step 198.

The process in accordance with one embodiment of the invention of restoring lost

files to the remote computer is illustrated by the flow chart of Figs. 7A-7C. A remote

computer user initiates a restoration process, step 210, by either running a restoration program on the remote computer, by connecting to the central computer and accessing a

restoration program, or by calling the central computer administrator.

If the process is initiated on the remote computer, the restoration program searches

the FI records to retrieve all date/time stamps in which backups were performed, step 212.

The backup dates are displayed to the user, step 214, and the user selects the desired date

from the list, step 216. The restoration program on the remote computer then retrieves

each FI record, step 218, and checks whether the date/time stamp predates (meaning, as

used herein, is before or equal to) the selected backup date, step 220. If it does, the

program checks, step 222, for the presence of any deletion records which, in the preferred

embodiment, are FI records having duplicate file descriptors and file signatures as the

current FI record but an action code of "erased." In some embodiments, these deletion

records are located or indexed in the FI list to immediately follow the original FI record,

as explained above. If any deletion record exists and has a date/time stamp which

predates the user-selected backup date, step 222, a FI backup flag is not set, and, if there

are more FI records to process, step 226, the next FI record is retrieved, step 218.

If there is no deletion record for the current FI record, or if any deletion record

does not predate the backup date, a FI backup flag is set, step 224, indicating that the FI

record corresponds to a file to be restored. If there are more FI records to process, step

226, the next FI record is retrieved, step 218 and the process repeated until all FI records

have been checked. If the restoration process is performed on the central computer, either by the

remote computer user remotely accessing the central computer restoration program or by

the central computer administrator, a similar checking process is performed with records

in the FOL and FFL. All FOL records with a remote computer ID corresponding to the

particular remote computer are retrieved and the date/time stamps checked to obtain a list

of dates. The user selects the desired backup date, and FOL records which predate that

backup date are retrieved. The FFL records having FFL File IDs matching those in the

retrieved FOL records are retrieved and, accounting for deletion records, are used to

obtain a list of files.

If the remote computer is used and the files are to be transmitted on-line or over a

network or other communication link, step 230 (Fig. 7B), a list of the files corresponding

to the FI records with backup flags set is displayed to the user, step 232. The user selects

the desired files to be restored, step 234, and the selected FI records are stored in a

restoration requested database, step 236. The database is transmitted to the central

computer, step 238, which receives the database, opens it, step 240, and retrieves the

compressed files from backup storage 16, step 242. The central computer transmits the

retrieved compressed files to the remote computer, step 244, and the files are

decompressed at the remote computer, step 246 and placed in their proper location (drive

and directory) on the remote computer. The restoration process is thus complete.

If the files are not to be transmitted from the central computer (Fig. 7C), the FI

records, either transmitted from the remote computer or obtained from the FOL and FFL as explained above, are stored in a restoration database, step 250, and the central

computer retrieves the compressed files corresponding to the FI records from backup

storage 16, step 252. The retrieved compressed files are stored on a high capacity storage

medium such as a CD-ROM, step 256, and the storage medium is sent to the remote

computer user, step 256. The files from the storage medium are then loaded into the

remote computer, step 258, and decompressed, step 260.

Figs. 8-15 contain screen displays used in one implementation of the present

invention in a software program developed by the assignee of the present application,

Telebackup Systems Inc. of Calgary, Alberta. The program is designed to operate with

the Windows95 operating system available from Microsoft Corporation of Redmond,

Washington. Fig. 8 shows a configuration screen which allows for the entry of a user

name and organization and for the selection of a particular modem and modem

parameters. Fig. 9 shows a billing screen by which the central computer administrator

tracks users for billing purposes.

Fig. 10 shows an installation screen in which a user is assigned a password. The

user must employ the password to gain access to its files backed up on the central

computer. Fig. 11 shows a backup scheduling screen in which the remote computer user

can specify times in which backup is to be performed and times when backup should not

be performed. Fig. 12 shows a backup configuration dialog box in which the user can

specify which drives to backup. Fig. 13 shows a file exclusion dialog box in which the

user can specify which files should not be backed up. Fig. 14 shows a file restoration dialog box in which the user is presented with the

dates available for backup and, once a backup date is selected, the files available to be

restored for that date. The user can also select whether the files should be sent by modem

transmission or stored on a CD-ROM. Fig. 15 shows a restoration status dialog box

showing the progress of an on-line file restoration.

While the invention has been described and illustrated in connection with preferred

embodiments, many variations and modifications as will be evident to those skilled in this

art may be made without departing from the spirit and scope of the invention. The

invention is thus not to be limited to the precise details of methodology or construction

set forth above as such variations and modification are intended to be included within the

scope of the invention.

Claims

WHAT IS CLAIMED IS:

1. In a data processing system comprising a plurality of remote computers which

are connectable to a central computer, a method for backing up files stored on the remote

computers to the central computer, comprising:

storing a file list in a memory device accessible to the central computer, the file list

containing a plurality of records each record corresponding to a file stored at some time

on one or more of the remote computers, each record comprising file identification data

identifying a respective file and including a signature of the respective file;

receiving at the central computer from a first remote computer first file

identification data which identifies a first file stored at some time on the remote computer

and which includes a signature of the first file;

comparing the first file identification data including the first file signature received

from the first remote computer with the file identification data of one or more records

contained in the file list; and

if the first file identification data received from the first remote computer does not

match the file identification data of any of the records contained in the file list,

transmitting a message from the central computer to the first remote computer instructing

the first remote computer to transmit the first file to the central computer.

2. The method of claim 1 comprising, if the first file identification data received

from the first remote computer does not match the file identification data of any of the records contained in the file list, adding a record containing the first file identification

data to the file list.

3. The method of claim 1 comprising, for each of a plurality of files stored at

some time at the first remote computer, determining whether the file is:

a new file which has not previously been backed up from the remote computer to

the central computer;

a changed file for which an earlier version thereof has previously been backed up

from the remote computer to the central computer; or

a deleted file which has previously been backed up from the remote computer to

the central computer and which has since been deleted from the first remote computer,

and wherein the first file identification data received at the central computer

identifies a new, changed, or deleted file.

4. The method of claim 3 comprising receiving at the central computer status data

associated with the first file identification data, the status data indicating whether the first

file identification data identifies a new, changed, or deleted first remote computer file.

5. The method of claim 4 comprising, if the status data indicates that the first file

identification data identifies a new or changed file, and if the first file identification data

received from the first remote computer matches file identification data of a record

contained in the file list, storing an indication that the first file identified by the first file

identification data is stored on the first remote computer.

6. The method of claim 5 comprising storing a file ownership list in a memory

device accessible to the central computer, the file ownership list containing a plurality of

records each of which corresponds to a file stored on one remote computer, each record

containing a remote computer identifier uniquely identifying one of the plurality of

remote computers and a pointer to a file list record, and wherein the step of storing an

indication that the first file identified by the first file identification data is stored on the

first remote computer comprises adding a record to the file ownership list containing an

identifier for the first remote computer and a pointer to the matching file list record.

7. The method of claim 6 wherein each record in the file list has a unique file list

identifier, and wherein the step of adding a record to the file ownership list containing a

pointer comprises adding a record containing the file list identifier of the matching file list

record.

8. The method of claim 5 comprising, if the status data indicates that the first file

received from the first remote computer matches the file identification data of a record

contained in the file list, transmitting a message from the central computer to the first

computer indicating that the first file need not be transmitted to the central computer.

9. The method of claim 4 comprising, if the status data indicates that the first file

identification data identifies a deleted file, storing an indication that the first file identified

by the first file identification data has been deleted from the first remote computer.

10. The method of claim 9 comprising, if the status data indicates that the first file

identification data identifies a deleted file, the step of locating a record in the file list

containing file identification data which matches the first file identification data, and

wherein the step of storing an indication that the first file identified by the first file

identification data has been deleted from the first remote computer comprises adding a

record to the file list which is identical to the located matching file list record.

1 1. The method of claim 1 comprising:

storing on the first remote computer an image file containing a plurality of records

each of which corresponds to a file stored at some time on the first remote computer and

backed up on the central computer, each image file record containing (i) identification

data identifying a respective file including a signature of such file and (ii) an indicator of

whether the respective file has previously been backed up on the central computer;

for each of a plurality of files stored on the remote computer, comparing respective

file identification data for that file with the file identification data of one or more image

file records.

12. The method of claim 1 wherein the step of receiving first file identification

data at the central computer comprises receiving first file identification data which

identifies a plurality of files stored at some time on the first remote computer.

13. The method of claim 12 comprising, for each of the plurality of files identified

by the first file identification data, determining whether the file is: a new file which has not previously been transmitted from the first remote

computer to the central computer;

a changed file for which an earlier version thereof has previously been transmitted

from the first remote computer backed up to the central computer; or

a deleted file which has previously been transmitted to the central computer from

the first remote computer and which has since been deleted from the first remote

computer; and

performing the step of comparing first file identification data to file identification

data contained in the file list only with respect to new, changed, or deleted files.

14. The method of claim 1 comprising storing a file ownership list in a memory

records each of which corresponds to a file stored on one of the plurality of remote

computers, each record containing a remote computer identifier uniquely identifying one

of the plurality of remote computers and a pointer to a given file list record.

15. The method of claim 14 comprising, upon a request to restore all or some of

the files stored on the first remote computer, the steps of:

retrieving records from the file ownership list having remote computer identifiers

corresponding to the first remote computer;

retrieving file list records pointed to by the pointers in the retrieved file ownership

file records; and transmitting the files identified by the file identification data contained in the

retrieved file list records from the central computer to the first remote computer.

16. The method of claim 15 wherein the file ownership list records each contain a

stamp indicating the date or time a file identified by the file identification data contained

in the file list record pointed to by the pointer was received by the central computer or

was indicated as deleted from the identified remote computer, and wherein only some of

the files stored on the first remote computer are to be restored, comprising specifying a

date or time of the files to be restored and selecting the files to be restored from among

the file ownership list records.

¹7. The method of claim 1 comprising, if the first file identification data received

from the first remote computer matches the file identification data of at least one of the

records contained in the file list, transmitting a message from the central computer to the

first remote computer indicating that the first file need not be transmitted to the central

computer.

18. The method of claim 1 comprising:

storing a counter associated with each record in the file list; and

if the first file identification data received from the first remote computer matches

the file identification data of at least one of the records contained in the file list,

incrementing the counter associated with the record containing the matching file

identification data.