File system for a plurality of storage classes

• US 4,993,030 A
• Filed: 04/22/1988
• Issued: 02/12/1991
• Est. Priority Date: 04/22/1988
• Status: Expired due to Term

• Current Assignee: Amdahl Corporation (Fujitsu Limited)
• Original Assignee: Amdahl Corporation (Fujitsu Limited)
• Inventors: Colgrove, John A., Gawlick, Dieter, Krakauer, Arno S., Wilmot, Richard B. II
• Primary Examiner(s): Shaw, Gareth D.
• Assistant Examiner(s): Ray, Gopal C.
• Application Number:
  US07/185,179
• Time in Patent Office: 1,026 Days
• Field of Search: 364/200, 364/900, 371/51.1, 371/40.1
• US Class Current: 714/6.12
• CPC Class Codes: G06F 11/1008 in individual solid state devices G06F11/1004 takes precedence
• CPC Class Codes: G06F 11/201 between storage system components
• CPC Class Codes: G06F 11/2056 by mirroring
• CPC Class Codes: G06F 11/2089 Redundant storage control functionality
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• CPC Class Codes: G06F 12/08 in hierarchically structured memory systems, e.g. virtual memory systems
• CPC Class Codes: G06F 15/161 Computing infrastructure, e.g. computer clusters, blade chassis or hardware partitioning casings, cabinets, racks or drawers for data centers H05K5/00
• CPC Class Codes: G06F 16/10 File systems; File servers
• CPC Class Codes: G06F 9/50 Allocation of resources, e.g. of the central processing unit [CPU]
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

A file system for managing data files for access by a plurality of users of a data processing system that includes internal storage for buffering, external storage, and a file user interface by which the plurality of users request access to data files. A first level, coupled to the file user interface in the internal storage allocates the internal storage for temporary storage of data to be accessed by the plurality of users, and generates requests for transactions with external storage in support of such allocations. A second level is coupled to the first level and the external storage and responds to the request for transactions with the external storage for managing the transactions for storage of data to, and retreival of data from, the external storage. The second level defines a plurality of physical storage classes which are characterized by pre-specified parameters that allocate data files subject of transactions to locations in external memory. In response to requests for transactions, the second level identifies one of the plurality of storage classes assigned to the data file subject of the transaction and carries out the transaction with the appropriate locations in external memory. At least one of the plurality of storage classes provides for utilization of a plurality of access paths in parallel for transactions involving data files assigned to such storage class. Further, at least one parameter pre-specified for storage classes identifies a level of reliability desired for the subject data files.

Claims

1. An apparatus for controlling accesses to data in data files by a plurality of users of a data processing system, the data processing system including internal storage, external storage storing data files, a plurality of access paths between internal storage and external storage, and a file user interface by which the plurality of users request access to data files, the apparatus comprising:

first means, coupled to the file user interface and the internal storage, for allocating the internal storage for temporary storage of current data to be accessed by the plurality of users, for deallocating internal storage from old data stored in the internal storage, and for generating requests for accesses with external storage for transfer of the current data to the allocated internal storage, and of updated old data from the deallocated internal storage; and

second means, coupled to the first means and the external storage and responsive to the requests for accesses, for managing transfers between internal storage and external storage through the plurality of access paths.

2. The apparatus of claim 1, wherein the second means includes:

logical mapping means, responsive to the requests for accesses to external storage, for assigning a logical address to data subject of each access, and

external mapping means, responsive to the logical address, for mapping the logical address to external storage and carrying out the access subject of the request with external storage.

3. The apparatus of claim 2, wherein there is a plurality of physical storage classes characterized by prespecified parameters that map the logical address of a data file subject of the access to locations in external memory, and the external mapping means includes:

means, responsive to the logical address, for identifying one of the plurality of storage classes; and

means, responsive to the identified storage class, for carrying out the access with the locations in external memory to which the prespecified parameters map data files in the identified storage class.

4. The apparatus of claim 3, wherein the second means further includes:

means, responsive to the identified storage class, for generating error codes for data subject of an access for transfer of data from internal storage to external storage; and

means, responsive to the identified storage class, for detecting and correcting errors in data subject of an access for transfer of data from external storage to internal storage.

5. The apparatus of claim 3, wherein storage classes in a subset of the plurality of storage classes are characterized by parameters specifying a cell of data, the cell including a plurality of local cells that may be accessed across a plurality of access paths in parallel, a cell of data being specified by a first parameter W defining a number of access paths to corresponding local cells of data for parallel access to a cell, wherein W local cells define a cell, and a second parameter D defining a number of blocks of data from the data files in the storage class within each local cell.

6. The apparatus of claim 5, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and there is at least one block of data in each local cell.

7. The apparatus of claim 5, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and storage classes in a subset of the plurality of storage classes are each further characterized by a reliability parameter that specifies an error recovery algorithm, and the external mapping means further includes:

means, responsive to the reliability parameter, for generating error codes for data subject of an access for transfer of data from internal storage to external storage; and

means, responsive to the reliability parameters, for detecting and correcting errors in data subject of an access for transfer of data from external storage to internal storage.

8. The apparatus of claim 5, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and storage classes in a subset of the plurality of storage classes are each further characterized by a reliability parameter that specifies one of a plurality of error recovery algorithms, and the external mapping means further includes:

means, responsive to the reliability parameter, for implementing the error recovery algorithm.

9. The apparatus of claim 8, wherein one of the plurality of error recovery algorithms provides for replication of local cells of data subject of an access for transfer of data from internal storage to external storage, and for storage of replicated local cells in external storage across independent access paths in parallel, and for selection of a best one of replicated local cells of data subject of an access for transfer of data from external storage to internal storage.

10. The apparatus of claim 8, wherein one of the plurality of error recovery algorithms provides for generation of an error code for each cell, and storage of the error code to a local cell within the cell in external storage in parallel with storage of the data subject of an access for transfer of data from internal storage to external storage;

and for detection and correction of errors in data subject of an access for transfer of data from external storage to internal storage.

11. The apparatus of claim 8, wherein the error code comprises bitwise parity over the local cells of data within the cell.

12. The apparatus claim 8, wherein for a local cell including N bits, the error code comprises N×

N bit code, where M is a number greater than one, stored in M local cells within the cell.

13. The apparatus of claim 1, wherein the apparatus further includes:

means, coupled to the first means, for specifying dependencies by users on locations in the internal storage allocated to the plurality of users and wherein said first means is responsive to the specified dependencies in allocating and deallocating the internal storage.

14. The apparatus of claim 13, wherein the dependencies specified by the means for specifying are programmable through the data processing system for global balancing of the allocation and deallocation of internal storage for the plurality of users.

15. The apparatus of claim 13, wherein the first means includes:

means, responsive to the dependencies, for locating, journaling and releasing locations to allocate and deallocate the locations in internal storage for the plurality of users.

16. The apparatus of claim 1, wherein there is a plurality of physical storage classes characterized by prespecified parameters that map data files subject of accesses to locations in external memory, and the second means includes:

means, responsive to requests for accesses, for identifying one of the plurality of storage classes assigned to a data file subject of an access; and

means, responsive to the identified storage class, for carrying out the access with the locations in external memory to which the prespecified parameters map data files in the identified storage class.

17. The apparatus of claim 16, further including:

means, coupled to the second means, for assigning data files to storage classes.

18. The apparatus of claim 17, further including means for programming the means for assigning for dynamic allocation of external memory.

19. The apparatus of claim 1, wherein the second means includes means, responsive to a single request for access with external storage, for carrying out the access subject of the request across a subset of the plurality of access paths in parallel.

20. The apparatus of claim 1, wherein there is a plurality of physical storage classes characterized by prespecified parameters that map the data file subject of the access to locations in external memory, and the second means includes:

means, responsive to a request for access with external storage, for identifying one of the plurality of storage classes assigned to a data file subject of request for access; and

means, responsive to an identified storage class that maps a data file to a plurality of locations accessible through a subset of the plurality of access paths, for carrying out the access through the subset of the plurality of access paths in parallel.

21. The apparatus of claim 1, wherein there is a plurality of the first means and a plurality of the second means, to support continuous operation in the event of a single point of failure.

22. The apparatus of claim 1, wherein the external storage includes a plurality of physical storage devices, and the second means includes:

means, independent of the first means, for carrying out transfer of data from data files stored in a first device to a second device in the plurality of physical storage devices.

23. The apparatus of claim 20, wherein the second means further includes:

means, independent of the first means, for carrying out transfer of data in data files assigned to a first storage class from locations identified in the first storage class to locations identified in a second storage class in the plurality of storage classes.

24. An apparatus for controlling access to data in data files by a plurality of users of a data processing system, the data processing system including internal storage, external storage, a plurality of access paths between internal storage and external storage, and a file user interface by which the plurality of users requests access to data files, wherein there is a plurality of physical storage classes characterized by prespecified parameters that map a data file subject of an access to locations in external memory, the apparatus comprising:

first means, coupled to the file user interface and the internal storage, for allocating the internal storage for temporary storage of current data in the data files to be accessed by the plurality of users, for deallocating internal storage from old data stored in the internal storage, and for generating requests for accesses with external storage for transfer of current data to allocated internal storage, and of updated old data from deallocated internal storage; and

means, coupled to the first means, for specifying dependencies by users on locations in the internal storage allocated to the plurality of users, wherein the first means is responsive to the specified dependencies in allocating and deallocating the internal storage;

logical mapping means, responsive to the requests for accesses with external storage, for assigning a logical address to data subject of each access;

external mapping means, responsive to the logical address, for identifying one of the plurality of storage classes; and

means, responsive to the identified storage class, for carrying out the access with the appropriate locations in external memory through a subset of the plurality of access paths.

25. The apparatus of claim 24, wherein the first means includes:

means, responsive to the dependencies, for locating, journaling and releasing locations to allocate and deallocate the locations in internal storage for the plurality of users.

26. The apparatus of claim 24, wherein the dependencies specified by the means for specifying are programmable through the data processing system for global balancing of the allocation and deallocation of internal storage for the plurality of users.

27. The apparatus of claim 24, wherein the means for carrying out the access is responsive to a single request for access with external storage to transfer data across a subset of the plurality of access paths in parallel.

28. The apparatus of claim 27, wherein storage classes in a subset of the plurality of storage classes are characterized by parameters specifying a cell of data, the cell including a plurality of local cells that may be accessed across a plurality of access paths in parallel, a cell of data being specified by a first parameter W defining a number of access paths to corresponding local cells of data for parallel access to a cell, wherein W local cells define a cell, and a second parameter D defining a number of blocks of data in the data files in the storage class within each local cell.

29. The apparatus of claim 28, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and there is at least one block of data in each local cell.

30. The apparatus of claim 24, wherein the means for carrying out accesses with external storage further includes:

means, responsive to the identified storage class, for generating error codes for data subject of an access for transfer of data from internal storage to external storage; and

means, responsive to the identified storage class, for detecting and correcting errors in data subject of an access for transfer of data from external storage to internal storage.

31. The apparatus of claim 28, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and storage classes in a subset of the plurality of storage classes are each further characterized by a reliability parameter that specifies an error recovery algorithm, and the means for carrying out accesses with external storage further includes:

means, responsive to the reliability parameter, for generating error codes for data subject of an access for transfer of data from internal storage to external storage; and

means, responsive to the reliability parameter, for detecting and correcting errors in data subject of an access for transfer of data from external storage to internal storage.

32. The apparatus of claim 28, wherein a block of data is a basic unit of transfer across an access path between external storage and internal storage, and storage classes in a subset of the plurality of storage classes are each further characterized by a reliability parameter that specifies one of a plurality of error recovery algorithms, and the means for carrying out accesses with external storage further includes:

means, responsive to the reliability parameter, for implementing the error recovery algorithm.

33. The apparatus of claim 32, wherein one of the plurality of error recovery algorithms provides for replication of local cells of data subject of an access for transfer of data from internal storage to external storage, and for storage of replicated local cells in external storage across independent access paths in parallel, and for selection of a best one of replicated local cells of data subject of an access for transfer of data from external storage to internal storage.

34. The apparatus of claim 32, wherein one of the plurality of error recovery algorithms provides for generation of an error code for each cell, and storage of the error code to a local cell within the cell in external storage in parallel with storage of the data subject of an access for transfer of data from internal storage to external storage;

and for detection and correction of errors in data subject of an access for transfer of data from external storage to internal storage.

35. The apparatus of claim 32, wherein the error code comprises bitwise parity over the local cells of data within the cell.

36. The apparatus of claim 32, wherein for a local cell including N bits, the error code comprises a N×

N bit code, where M is a number greater than one, stored in M local cells within the cell.

37. An apparatus in communication with a data processing system for storing a data file for sequential transfers to the data processing system, the data file including a plurality of local cells, each local cell including at least one block of data, wherein a block is a basic unit of transfer between the data processing system and the apparatus, the apparatus comprising:

a plurality of storage means for storing data;

a plurality of logical input/output paths P_n, for n equal to 1 through N, each path coupled to a subset of the plurality of storage means and to the data processing system, so that local cells of data may be transmitted in parallel through the N paths between the data processing system and the plurality of storage means; and

wherein

the local cells in the data file are stored in the plurality of storage means in a sequence of local cells LC_i for i equal to 1 to X, and wherein local cell LC_i is stored in a storage P_n+k means coupled to path P_n ; and

local cell LC_i+l is stored in a storage means coupled to path P_n+k, where k is not equal to zero.

38. The apparatus of claim 37, wherein the data file includes S cells of data, each cells including W local cells, where W is less than or equal to N and S equals X/W rounded to the next higher integer, each cell including at least one local cell storing an error correction code for the cell, and all local cells in a given cell stored in storage means coupled to different paths.

39. The apparatus of claim 38, wherein the error correction code for a given cell comprises a bitwise exclusive-OR of all local cells in the cell, except the error correction code.

40. The apparatus of claim 39, wherein the local cells in the data file containing the error correction codes are stored in storage means coupled to a single path.

41. The apparatus of claim 39, wherein the local cell in the data file containing the error correction code for cell j, is stored in a storage means coupled to path P(((j-1)modW)+1).

42. The apparatus of claim 38, wherein each local cell includes L bits, and the error correction code comprises a L×

M bit code, where M is an integer greater than one.

43. An apparatus in communication with a data processing system for storing a data file for sequential transfers to the data processing system, the data file including a plurality of local cells, each local cell including at least one block of data that can be manipulated by the data processing system as a unit for access to the data file, the apparatus comprising:

a plurality of storage means for storing data;

a number W of logical input/output paths, path 1 through path W, each path coupled to a subset of the plurality of storage means and to the data processing system, so that blocks of data may be transmitted in parallel through the W paths between the data processing system and the plurality of storage means; and

wherein

the data file is stored in the plurality of storage means in a sequence of X local cells, and local cell 1 in the sequence is stored in a storage means coupled to path 1, local cell 2 is stored in a storage means coupled to path 2, local cell W is stored in a storage means coupled to Path W, local cell W+1 is stored in a storage means coupled to path 1, local cell W+2 is stored in a storage means coupled to path 2, local cell 2W is coupled to a storage means coupled to path W, and local cell X is stored in a storage means coupled to path (((X-1)modW)+1).

44. The apparatus of claim 43, wherein the data file includes S cells of data, each cell including up to W local cells, where S equals X/W rounded to the next higher integer, each cell including at least one local cell storing an error correction code for the cell.

45. The apparatus of claim 44, wherein the error correction code for a given cell comprises a bitwise exclusive-OR of all local cells in the cell, except, for the local cell storing the error correction code.

46. The apparatus of claim 45, wherein the local cells in the data file containing the error correction codes are stored in storage means coupled to a single path.

47. The apparatus of claim 45, wherein the local cell in the data file containing the error correction code for cell j, is stored in a storage means coupled to path P(((j-1)modW)+1).

48. The apparatus of claim 44, wherein each local cell includes L bits, and the error correction code comprises a L×

M bit code, where M is an integer greater than one.

49. An apparatus in communication with a data processing system for storing a data file for sequential transfers to the data processing system, the data file including a sequence of local cells LC_i for i equal to 1 through X, each local cell including at least one block of data that can be manipulated as a unit by the data processing system for access to the data file, the apparatus comprising:

a plurality of storage means for storing data;

a plurality of logical input/output paths P_n, n equal to 1 through W, each path coupled to a subset of the plurality of storage means and to the data processing system, so that blocks of data may be transmitted in parallel through the plurality of paths between the data processing system and the plurality of storage means; and

means, coupled to the plurality of paths, for allocating the sequence of local cells LC_i, for i equal to 1 to X, to at least a subset of the plurality of storage means, so that local cell LC_i is stored in a storage means coupled to path P_n, local cell LC_i+1 is stored in a storage means coupled to path P_n+k P_k, where k is not equal to zero.

50. The apparatus of claim 49, wherein local cell LC_i is stored in a storage means coupled to path P_n, and n is equal to (((i-1)modW)+1), for i equal to 1 to X.

51. An apparatus in communication with a data processing system for storing a data file for sequential transfers to the data processing system, the data file including a sequence of local cells LC_i, for i equal to 1 through X, each local cell including at least one block of data that can be manipulated as a unit through the data processing system by users of the data file, the apparatus comprising:

a plurality of storage means for storing data;

a plurality of logical input/output paths P_n, for n equal to 1 through W, each path coupled to a subset of the plurality of storage means and to the data processing system, so that blocks of data may be transmitted in parallel through the plurality of paths between the data processing system and the plurality of storage means;

means, coupled to the plurality and paths to receive the blocks of data, for generating an error correction code ECC_s for each set s of E local cells, where s goes from 1 to S, and S is equal to X/E rounded to the next larger integer; and

means, coupled to the plurality of paths and to the means for generating an error correction code ECC_s, for mapping the sequence of local cells LC_i, for i equal to 1 to X, and error correction codes ECC_s, for s equal to 1 to S, to at least a subset of the plurality of storage means, so that all local cells in the set s, and error correction code ECC_s, define a cell, and local cells in the cell and each are stored in storage means coupled to different paths.

52. The apparatus of claim 51, wherein the error correction codes ECC_s are generated by taking the bitwise exclusive-OR of all local cells of data in the cell.

53. The apparatus of claim 52, wherein the means for mapping maps all error correction codes to storage means coupled to a single path.

54. The apparatus of claim 52, wherein the means for mapping maps error correction code ECC_s to a storage means coupled to path P.sub.(((s-1)modW)+1)

55. The apparatus of claim 51, wherein the error correction codes are multiple bit Hamming codes.

56. The apparatus of claim 51, wherein the error correction code ECC_s has a size equal to M local cells, and E is equal to W minus M.

57. An apparatus in communication with a data processing system for storing a plurality of data files for sequential transfers to the data processing system, each data file including a sequence of local cells LC_i, for i equal to 1 through X, where X is a parameter associated with each data file, each local cell including at least one block of data that can be manipulated as a unit through the data processing system by users of the data file, the apparatus comprising:

a plurality of storage means for storing data;

a plurality of logical input/output paths P_n, n equal to 1 through N, each path coupled to a subset of the plurality of storage means and to the data processing system, so that data may be transmitted in parallel through the plurality of paths between the data processing system and the plurality of storage means; and

means, coupled to the plurality of paths and to the data processing system, and responsive to parameters associated with each data file by the data processing system, for mapping the sequence of local cells LC_i, for i equal to 1 to X, for each data file to at least a subset of the plurality of storage means, so that local cell LC_i of a given data file is stored in a storage means coupled to path P_n, and local cell LC_i+1 of the given data file is stored in a storage means coupled to path P_k, where k is not equal to n.

58. The apparatus of claim 57, wherein the number of blocks per local cell is an additional parameter associated with each data file by the data processing system, and to which the means for mapping is responsive.

59. The apparatus of claim 57, wherein a number W of local cells equal a cell, and the number W is an additional parameter associated with each data file by the data processing system.

60. The apparatus of claim 59, wherein local cell LC_i is stored in a storage means coupled to P_n, and n is equal to (((i-1)mod W)+1), for i equal to 1 to X.

61. An apparatus in communication with a data processing system for storing a plurality of data files for sequential transfers to the data processing system, each data file including a sequence of local cells LC_i, for i equal to 1 through X, where X is a parameter associated with each data file, N local cells equal a cell, and each local cell includes at least one block of data that can be manipulated as a unit through the data processing system by users of the data file, the apparatus comprising:

a plurality of storage means for storing data;

a plurality of logical input/output paths P_n, n equal to 1 through N, each path coupled to a subset of the plurality of storage means and to the data processing system, so that data may be transmitted in parallel through the plurality of paths between the data processing system and the plurality of storage means;

means, coupled to the plurality of paths and the data processing system to receive the blocks of data, for generating and error correction code ECC_s, for storage in Z local cells, for a set s of E local cells of data, where s goes from 1 to S, and S equal to X/E rounded to the next larger integer, and N is equal to Z+E; and

means, coupled to the plurality of paths and to the data processing system, and responsive to the parameters associated with each data file by the data processing system, for mapping the sequence of local cells LC_i, for i equal to 1 to X, and the error correction codes ECC_s, for s equal to 1 to S, for each data file to a subset of the plurality of storage means, so that all local cells of data in a given set and the local cells or error correction codes for the given set equal a cell of W local cells and each local cell in a cell is stored in storage means coupled to different paths.

62. The apparatus of claim 61, wherein the number of blocks per local cell is an additional parameter associated with each data file by the data processing system, and to which the means for mapping is responsive.

63. The apparatus of claim 61, wherein W is a parameter associated with each data file by the data processing system.

64. The apparatus of claim 61, wherein there is a plurality of types or error correction codes, and the type of error correction code is a parameter associated with each data file by the data processing system.

65. The apparatus of claim 64, wherein a first type of error correction code is generated by taking the bitwise exclusive-OR of all local cells in the set.

66. The apparatus of claim 64, wherein a first type of error correction code is generated by taking the bitwise exclusive-OR of all local cells in the set, and the means for mapping maps all error correction codes to storage means coupled to a single path.

67. The apparatus of claim 64, wherein a second type of error correction code is generated by taking the bitwise exclusive-OR or all local cells in the set, and the means for mapping maps error correction code ECC_s to a storage means coupled to path P.sub.(((s-1)modW)+1).

68. The apparatus of claim 64, wherein one type of the error correction codes is a multiple bit Hamming code generated over all local cells in the set.