File index system for mass storage device

Abstract

An optical archive storage system includes a large number of off-line optical disks which are selectively transported to read/write units by a robotic manipulator. Access to the data files on the disks is maintained and controlled through the use of a token file on a magnetic storage device on a controller computer for each data file on each disk. The file attributes which the operating system of the computer assigns to the token file are used to control access and ownership of the data files on the disks.

Description

FIELD OF THE INVENTION

The present invention relates to the organization and indexing of files in general, and relates, in particular, to the easy and convenient indexing and retrieval of files from mass storage devices in which the files themselves may not be on-line.

BACKGROUND OF THE INVENTION

There exists many forms of data storage devices for use with digital computers. It has now become appropriate, for some applications which require particularly intensive use of storage capability, to design mass storage devices which can maintain and store indefinitely large quantities of data. One form in which such a mass storage device can be implemented is an optical disk system. Commercially available write-once, read-many (WORM) systems are available. For example a two sided 30 centimeter diameter glass optical disk of the type developed by the Philips Corporation can be read and write on a Lazer Magnetic Storage International 1200 Read/Write Unit, with each WORM optical disk having a storage capability of 1 gigabyte (2.sup.30) bytes of data.

The overall storage capacity of an optical reader based mass storage device can be improved by using many optical disks with a small number of units for reading and writing on the disks. Unfortunately, the use of such a strategy means that the information on most of the disks is, at all times, off-line. This complicates the user interface to the disks and also makes more complicated functions which a user may normally expect, such as access to disk directories and searches of various file attributes.

SUMMARY OF THE INVENTION

The present invention is summarized in that a mass storage device includes a library of discrete file storage units, at least one reader to read the file storage units and a digital computer to control access to the file storage units, the digital computer having a magnetic memory and being programmed to create in that magnetic memory a token file for each file on each mass storage unit, each token file including specifications of the actual file on the mass storage device including file specification and name, the mass storage unit on which the file is located, the identification and location of the file on the mass storage device and the size of the file and the computer also being programmed so that the utility programs for transferring data to or from the mass storage device continually update the token files on the magnetic storage device so that at all times the token files on the magnetic storage device represent a complete and accurate catalog of the contents and status of all of the files in the mass storage devices whether or not the individual mass storage units are on-line.

It is an object of the present invention to provide an efficient and economical system for handling a mass storage device consisting of a large number of individual mass storage units such as optical disks.

It is another object of the present invention to provide an easy manner for gaining access to a large mass storage device and the files therein without all of the files having to be on-line at any one instant.

It is another object of the present invention to provide a mass optical archive system storage device in which users can gain access to the storage to archive and restore files without having to have any particular knowledge about the organization or structure of the files themselves within the mass storage unit.

Other objects, advantages, and features of the present invention will become apparent from the following specification when taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic illustration of the concept of an optical archive storage constructed in accordance with the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Shown in FIG. 1 is the principal physical manifestation of an optical archival storage system constructed in accordance with the present invention. The archival storage system of FIG. 1 includes a large number of two-sided optical disks 12 which are housed in a large cabinet 14. The disks 12 are each two-sided one gigabyte optical disk cartridges. The cabinet 14 includes a plurality of vertical rows of racks which are vertically arranged to receive a number of the optical disk cartridges 12. A number of read/write units 16 are located in the middle of the rack 14. The read/write units are capable of writing one time into each location in each optical disk carried in each optical disk cartridge and then reading many times from the same location. A robot, generally indicated at 18 is provided to retrieve a particular optical disk cartridge 12 and to transport it and insert it into an appropriate one of the four read and write units 16. The control of the entire unit as illustrated in FIG. 1, including both the indexing for the location and file structure in each of the optical disk cartridges, the date of transfer to and from each of the read/write units 16, and the operation of the robot 18, is all under the control of a separate digital computer, in this case a VAX 11/750, not shown in FIG. 1. The controlling digital computer has the responsibility of selecting an appropriate optical disk cartridge 12 for transportation by the robot 18 into an unused one of the read/write units 16. The controlling computer therefore must have a directory of the files on each of the optical disks and be able to properly select disk cartridges for transport to and from the read/write unit 16. It is also advantageous for the computer to have information about the files contained in each individual disk on an on-line basis so that indexing and searches for particular files can be conducted without the necessity for transporting each disk cartridge individually to the read/write units to find the contents thereof. A related objective is that the access to files in each of the optical disks must be convenient to the users, and in particular to casual users, so that extensive training or instruction is not required for users to make access of the optical archive storage.

These objectives are implemented in the optical archive store in accordance with the present invention through the use of a "token" file contained in the magnetic memory of the controlling computer. The attributes of the token file itself are determined by the operating system of the operating computer. In the case of a VAX computer as used within the embodiment of the present invention described here, the VAX computer utilizes an operating system known as VMS, for virtual memory system, which implements a standard file system known as RMS. Each of the token files on the magnetic memory of the VAX system would therefore be stored as an RMS file and have associated with it all of the normal overhead components for any other RMS file under the operating system of the VAX computer. Within the token file itself, there may be placed additional information specific to the data requirements of handling the optical disks in the optical archival store. Together, the relevant attributes of the controlling computer operating system and the additional data contained within the token file provide all of the information reasonably necessary to the users to learn about, search, and determine ownership and access rights to any given file contained on an optical disk which is not on line at a given instant. In this way, swift and efficient maintenance of information about files is maintained at all times. In addition, since many attributes of the token file are maintained in the context of the environment of the operating system of the controlling computer, many casual users who simply have a knowledge about the operating system of such a computer will be able to use the optical archive storage without the need for great additional knowledge or information.

Therefore, the purpose of the token data base system is to provide an index to all of the files located on the optical archive system. The index is intended to provide the following capabilities:

The index should provide fast lookup of a file when the name of the file or other primary key information is known. The index should provide search capability when the name of the file is not known. The optical archive system is an archival system, therefore it is expected that many years may elapse between the time a file is written and the time it is again needed. A user may be able to remember only bits and pieces of the information needed to retrieve the file. The index should provide security services, so that only authorized persons will have access to the index information or to the data. The index should provide facilities for auditing the number of files and size of each file stored, so that storage costs can be billed to the owner.

Within each optical disk of the optical archive system there is an index log. The index log serves the purpose of an index file on a conventional magnetic disk except that it is constructed as a log since erasures cannot be made to it. Thus, for example, file headers cannot be erased from the log but newer versions can be written thereon superseding older versions. A sequence number field in the index log indicates the most recent version of the file header. The file header format on the optical disk itself contains information about the actual files contained on that disk and contains all the appropriate information for the files which are there. This information is contained in the large index file which includes such information as the serial numbers, the logical names, the physical locations, and individual attributes (access, owner, etc.) of all of the files on the disk. It is a block of this information that is written into the token file located in the magnetic storage of the controlling computer.

The token data base consists of a single small RMS file stored on magnetic disk for each individual file stored on optical disk. The RMS file on magnetic disk is a "token" that points to the corresponding optical file written on optical disk. The token file creates an illusion that the optical file resides on magnetic disk rather than optical disk. A subset of the RMS attributes of the token file are interpreted by the optical archive storage system to be the attributes of the corresponding optical file. Additional information abort the optical file is contained as data within the token file.

The optical disks of the optical archive storage system are grouped into logical volumes. Each volume consists of one or more optical disks. For this purpose each side of a double-sided cartridge is considered a separate optical disk, but both sides are generally placed within the same volume. There can be an arbitrary number of optical disks in any given volume. Under each volume, the files stored on the OAS system are arranged in directories and subdirectories up to seven levels deep.

The complete token file specification consists of "store::volume:[directory]name.ext;version" where store is the name of the optical archive storage system (if multiple systems exist on the same network,) volume is the name of the logical volume, and the remainder is equivalent to an ordinary RMS file specification. The entire token file specification uniquely identifies a token file. From the point of view of the RMS operating system, a token file is in all respects an ordinary RMS sequential text file, usually less than 512 bytes in length. In this way the optical archive system can make use of all standard RMS facilities for manipulating the token files. This greatly reduces the amount of programming needed to create an on-line data base with equivalent capability, because the system utilizes the effort that has already been placed into the RMS file system and the RMS operating system. For example, the RMS directory command can be used to display directory information about the token files, and the RMS file ownership and protection facilities can be used to enforce file ownership and protection.

The attributes of an RMS file under the RMS operating system that are relevant to the optical archive storage system are:

File specification, of the form "store::volume:[directory]name.ext;version", which is the name of the file. This is an ordinary RMS file specification. The name of the token file is interpreted by the optical archive storage system as the name of the file on optical disk. This is the primary key into the token data base.

Creation date, of the form "dd-mmm-yyyy hh:mm:ss.cc", which is the date on which the file was created. The creation date of the token file is interpreted by the system as the date the optical file was created on optical disk.

Modification date, of the form "dd-mmm-yyyy hh:mm:ss.cc", which is the date on which the attributes of the file were last changed. The modification date of the token file is interpreted by the system as the modification date of the optical file on optical disk.

File protection, of the form "(system,owner,group,world)", which is the user identification code based protection mask of the file. This is an ordinary RMS file protection. It is interpreted as the protection of the file on optical disk for purpose of determining access rights to the optical file.

User identification code, of the form [group,id] that identifies the owner of the file. This is an ordinary RMS user identification code. It is interpreted as the ownership of the file on optical disk for the purpose of determining access rights, and for allocating storage charges to a particular user of the OAS system.

Additional information stored within the token file consists of:

Optical disk serial number, is the serial number of the optical disk on which the file is located. This uniquely identifies the optical disk and serves as the key for discovering any other attributes of the optical disk which are needed in order to process a file request. It also identifies which optical disk of a volume the file is actually written on.

Optical file id, is the file id number of the file on the optical disk. Each file on an optical disk has a unique optical file id. The serial number and the optical file id can be used in combination as a key to find information on the optical file that is cached on magnetic disk. Information on the most frequently used optical disks is cached on magnetic disk in order to speed up the process of mounting an optical disk in a read/write unit.

Optical file id address, is the logical address on the optical disk where the optical file header of the optical file is located. The optical file header contains all information about the optical file. The optical file id address is used to speed the process of looking up optical file information when the file is to be modified or restored to magnetic disk.

Size, is the size of the file on optical disk in units of 512 byte blocks. This information is of interest to the owner of the file, and is also used in charging for the storage costs of the optical file.

Original file specification, of the form "node::device:[directory]name.ext;version", is the original file specification of the magnetic disk file that was copied to optical disk to create the optical file. This is maintained on-line in order to assist the user in finding a file based upon components of the original name.

Description, is a textual description of the optical file. It is maintained on-line in order to assist the user in finding a file based upon key words within the description. The description is optional, but it is preferred that all users of the optical archive storage system include a description with each file. Thus the actual data portion of the token file will consistently be rather modest in length. Because a large portion of the attributes of the optical archive file which are necessary to be stored is stored as part of the overhead of the RMS operating system, the data portion of the RMS file needs contain only the relatively modest amount of information referred to above which is additive to the other overhead RMS information. Illustrated just below is an example of the data portion of a sample token file. B14024 FID=00000403, FID ADDR=000F560F 0000000504 blocks written This file was archived from d13:[APVOAS]PDP.SAV; 1 save set of rt-.DELTA.1 oas drive code. In the example shown above the first numeral, B14024, refers to the serial number of the optical disk on which the file is stored. The second value which begins with "FID=" is the file ID of the optical file on the particular optical disk identified by the serial number. The third parameter, identified by "FIDADDR=" refers to the address of the optical file 403 identified in the previous file ID parameter. This is the logical address of the optical file as referred to above. The next line contains the file size which is indicated to be 504 blocks. The last two lines are the comments or narrative description of the file and contained therein is the original file specification indicating where the file was archived from. This is the entire contents of a token file data portion. All the remaining other attributes of the file necessary for data manipulations, search, and handling, as described in further detail below, are all contained within the overhead portion of the RMS file organization and do not require specific manipulation or handling or organization of file structure to maintain proper organization of the optical archive storage system.

The addition of the relevant RMS file attributes and the data portion of each token file provides all of the necessary information for each file contained on the optical archive system to uniquely determine the name of each optical file, determine which optical disk any given optical file resides on, to determine the ownership and access rights to any given optical file, to quickly look up relevant access information concerning any optical file, to quickly determine where on the optical disk the optical file header is located, to search for a particular optical file based on the original name of the file, to search for any optical file based upon key words in a text description, and to generate charging information as to the storage charges accrued for the storage of any particular optical file. The token data base structure allows all of these operations to be performed without the need to load any optical disk whatsoever into any of the read/write units contained in the optical archive storage system. The speed of the entire system is facilitated and efficient search of the entire token data base is practical for any file based on some combination of information that a user may remember.

The amount of programming required to create and maintain the token data base is minimized by the fact that standard RMS and RMS facilities are available to perform all the necessary operations. The optical archive storage system software should automatically maintain the token data base as each optical file operation is performed. There are five primary optical file operations:

Archive. Copy a file from magnetic disk to optical disk. This operation automatically creates a token file for the corresponding optical file. Access rights to archive a file can be determined by the user's access rights to create a file in the RMS directory where the token file will be located.

Restore. Copy a file from magnetic disk to magnetic disk. This operation uses the token file to quickly locate the optical file on the optical disk. In all cases the user need not know which optical disk the optical file resides upon. It is sufficient to know the name of the optical file, which is identical to the name of the corresponding token file. Access rights to restore a file can be determined by the user's access rights to read the token file.

Change. Change the ownership, protection, or other attributes of an optical file. This does not affect the date stored within the optical file. The file receives a new optical file header. The attributes and/or data of the corresponding token file are also changed. Access rights to change an optical file can be determined by the user's access rights to "control" the token file.

Rename. Change the name of the optical file. This does not affect the data stored within the optical file. The file receives a new optical file header. The name of the corresponding token file is also changed. Access rights to rename an optical file can be determined by the user's access rights to "control" the token file. If the directory location of the optical file is to be changed as part of the rename operation, the user must also have write access to the destination token file directory.

Retire. Mark an optical file as deleted. This is the equivalent of deleting an optical file, since the data is not actually erased. The optical file receives a new optical file header. The corresponding token file is deleted. Access rights to delete an optical file can be determined by the user's access rights to delete the token file.

There are a number of desirable auxiliary operations that involve the use of the token data base:

Directory. Perform a directory search on selected portions of the token data base. This is done by the optical archive storage system software using the RMS directory command and other RMS facilities. Only directory information relevant to the optical archive storage system is displayed. Special handling of the data stored in the token file is performed so that a user may request the size, original name, description or other information about the optical file which is not directly available to the RMS directory command.

Search. Perform a fast search of selected portions of the token data base in order to locate an optical file based upon some known items of information such as keywords in a description or components of the original file name. This is done directly by the user using the RMS search command.

Create/directory and other directory manipulations. Create, modify or delete a directory in the token data base. This is done by the optical archive storage system software or directly by the user using the corresponding RMS facilities.

Accounting. Account for storage charges of each optical file. This is done by sequentially processing all token data base files to determine the number of optical files and total size of all optical files owned by each user. The token file ownership and the size data within the token file are used for this purpose.

Exceptional operations. A token file can be "temporarily" changed to allow different access rights to an optical file without actually making a change on optical media. This is useful for one time operations where the original attributes are to be restored when the operation is completed. It may be desirable for a user to allow another user to have temporary access to his or her files. If every such instance required giving an optical file a new optical file header (to make the change permanent), an unnecessary waste of optical disk space would occur. Temporary changes to the token files can be made using standard RMS commands.

Rebuild token files. In the event that token files are accidentally lost or corrupted, they can be rebuilt from information contained on the optical disk. The token data base is preferably protected against loss or corruption by ordinary magnetic tape backup of the magnetic disk(s) on which it resides. Backup may be done using the RMS backup command.

Renaming, binding and unbinding optical disk volumes. The "store::volume:" portion of the token file specification is not recorded on optical media. This allows a volume to be renamed. An optical disk can be added or removed from a volume. A volume can be moved from one optical archive storage system to another. All these operations can be done without changing anything actually written on the optical media. These operations are performed using standard RMS commands.

Thus, it is possible relatively simply to program an optical archive storage system control program which makes use principally of the overhead of the VMS and RMS operating system and which requires only a few routines and special programs to properly handle and make available all of the file attributes for the files on the optical archive system. Thus, a user may need to know as few as five commands to do all of the things necessary to archive a file to the optical archive system and retrieve it. A first operation would be required to begin communication with the program and control of the optical archive storage system. A second command would be necessary to archive a given file onto the optical archive system and to assign it a name on the system. This second command would implement the archive operation described above and automatically result in the creation of an appropriate token file. A third command would be a retrieve or restore command which would read a file from the optical archive storage system and restore it in a given location on a magnetic disk. A fourth command would be a directory which would allow the user to obtain a directory of the users files located on the optical archive storage system. This would make use of the directory operation referred to above and would not require the system to access any of the optical disks. Lastly, the user would have to know the exit command to leave communication with the optical archive system. The use of the token file system as described herein makes the implementation of such a simple communication protocol with a casual user both practical and relatively efficient to implement.

Attached hereto is Appendix A and incorporated herewith by reference is a source code listing of the routines and sub-routines necessary to implement the optical archive storage operations referred to above. Also contained are printouts of the source code listings for the operations as to checking user access for various file operations. All of these routines are implemented in Fortrand and make extensive use of the VMS routines contained in the overheaded operating system implemented on a VAX computer of the type used for the implementation of the present invention by the inventors. One of ordinary skill in programming and familiar with the implementation of the VMS operating system may readily scan this code, which is fully commented, to understand in detail the operation of the functions described above and the manner in which those operations are implemented through the use of VMS utilities. One of ordinary skill in the art will also be able to ascertain from these code listings how to implement similar functions in other programming environments and with other machines.

In summary, the use of a token file data base as an index to access files contained on an optical archival storage system provides for a full range of indexing functions for a large mass storage device with a minimum of programming and maintenance. These objectives are accomplished by creating an illusion to the user that the optical file resides on a magnetic disk rather than on an optical disk so that the format and attributes of the file will be familiar to the casual user. Standard functions contained within the operating system of the controlling computer and the use of the utilities provided by that computer are utilized whenever possible. This allows for efficient searches of the index data base to the entire archival storage unit by searching the token data base and thereby avoiding the necessity for loading each individual optical disk onto a read/write unit during such searches. Thus the token data base is a strategy which reduces the cost of implementation of the optical archive storage unit and which enhances its index, particularly for very large storage units.

It is to be understood that the present invention is not limited to the particular arrangement and construction of parts as discussed above, but embraces all such modified forms thereof as come within the scope of the following claims. ##SPC1## ##SPC2## ##SPC3## ##SPC4## ##SPC5## ##SPC6## ##SPC7##

Claims

1. A mass storage system for digital data files comprising:

a plurality of individual storage units each containing at least one data file and an index log containing information about the at least one date file stored therein, each storage unit having an identification associated with it;

at least one reading device which can read the data files and index log of a storage unit placed therein;

automated means for placing a storage unit on the reading device; and

a digital computer having an associated magnetic storage device and connected to control the automated means to cause a one of the storage units to be placed on the reader, and connected to the reading device to receive data files therefrom, the computer being provided with an operating system program adapted for the maintenance, retrieval, and control of access to data files stored in the associated magnetic storage device, the computer programmed to maintain on the magnetic storage device a token file for each respective data file on each storage unit, each token file being maintained, retrieved and having access to it controlled by the operating system of the computer in accordance with the normal functioning of the operating system of the computer and each token file having, as its data, the identification of the storage unit on which the data file is placed, the address of the data file on the storage unit, and the size of the respective data file on the storage unit, whereby the file data which the operating system assigns to the token files is used to control access to the data files on the storage units without the need for access to the storage units.

2. A mass storage system as claimed in claim 1 wherein the storage units are optical disks.

3. A mass storage system as claimed in claim 1 wherein the automated means is a robot capable of transporting storage units under the control of the computer.

4. A mass storage system as claimed in claim 1 wherein the token file also includes a location into which a narrative description of the data file may be placed.

5. A mass storage system as claimed in claim 1 wherein the operating system of the computer normally maintains information about ownership, access and protection of data files on the associated magnetic storage device and wherein the computer is further programmed to determine ownership, access and protection of the data files on the storage units by referring only to ownership, access and protection attributes associated with the respective token file on the magnetic storage device for each data file.

6. A mass storage system as claimed in claim 1 wherein the computer is further programmed to create directories of data files by referring only to selected portions of the token files.

7. A mass storage system as claimed in claim 1 wherein the computer is a VAX computer and the token files are maintained by the VMS operating system.

8. An optical archive storage system for storing data files comprising:

a plurality of write-once, read-many times optical data disks stored off-line;

a read/write unit to read or write an optical disk mounted thereon;

a robotic manipulator to transfer optical disks into and out of the read/write unit;

a digital computer connected to control operation of the read/write unit and the robotic manipulator, the computer including an on-line magnetic disk storage device and containing an operating system adapted for the control of the access to and retrieval of the files on the on-line magnetic disk storage device, the on-line magnetic storage device having therein a token file for each data file on each optical disk, the token file including the identification of the data file, the identification of the optical disk, and identification of the subject matter of the data file so that directories and searches of the data files on the optical disks, and control of access to the files on the optical disks, can be performed using the token files without the need for access to the optical disks.

9. An optical archive storage system as claimed in claim 8 wherein the computer has an operating system and the token files are maintained by the operating system.

10. An optical archive storage system as claimed in claim 9 wherein the operating system assigns ownership, access and protection attributes to each token file and wherein the attributes of the token file are used to determine ownership, access and protection of the data files.