Patent EP0740247A2 successfully refreshed from Google Patents.
EP0740247A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Data stream server system
Assignee: HP Inc
•
Filing: 1996-04-19
•
Granted: —
•
Expires: 2016-04-19
Abstract
Abstract A server system (2) for storing and supplying data to a destination includes a plurality of disk drives (4) for storing data, a buffer memory (8) for temporarily storing a portion of the data from the disk drives (4), a reconstruction unit (10) for reconstructing a portion of the data from at least one of the disk drives (4) using a portion of the data from a plurality of the remaining disk drives (4) so that at least one of the disk drives (4) holding data belonging to the current parity group can be bypassed (either because that disk (4) is faulty or because there are already many queued read-requests for it), a control unit (12) for controlling the server system (2), and an output unit (14) for outputting data from either or both of the buffer memory (8) and the reconstruction unit (10). The server system (2) reduces congestion problems, reduces buffering requirements, and provides load balancing and fault tolerance. The server system may be a video storage server.
Inventors
- 👤 Yitzhak c/o Thaler Birk
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-11-02 20:41
CPC Classifications
G; G06; G06F; G06F11/00; G06F11/07; G06F11/08; G06F11/10; G06F11/1076; G06F11/1088; G11; G11B; G11B20/00; G11B20/10; G11B20/18; G11B20/1803; H; H04; H04N; H04N7/00; H04N7/16; H04N7/173; H04N7/17309; H04N7/17336
Claims Analysis
Total Claims: 15
Claim 1
Independent Claim
142 words
A server system for storing and supplying data, comprising: a plurality of memory units (4) each storing data; a buffer memory (8) operatively connected to said memory units, for temporarily storing a portion of data from said memory units; a reconstruction unit (10) operatively connected to said buffer memory for reconstructing a portion of data from at least one of said memory units using a portion of the data from a plurality of the remaining memory units so that at least one of said memory units can be bypassed; a control unit (12) operative to examine queues associated with said memory units to obtain queue information and then to decide on the basis thereof whether to bypass any of said memory units; and an output unit (14) for outputting data from either or both of said buffer memory and said reconstruction unit.
Claim 2
Dependent Claim
52 words
A server system as recited in claim 1, wherein data stored in said memory units (4) is partitioned into sets of data chunks. within each set the data chunks are equi-sized and each of the data chunks of a given set are stored on a different one of said memory units (4).
Claim 3
Dependent Claim
87 words
A server system as recited in claim 1 or 2, wherein the data is partitioned into sets of k equal-size data chunks, r redundancy chunks are created for each set of k data chunks, such that the data of any r of the k data chunks can be reconstructed from that of a plurality of remaining data chunks and redundancy chunks of the same set of chunks, and the k data chunks and the r redundancy chunks of each set are stored in k+r different memory units.
Claim 4
Dependent Claim
61 words
A server system as recited in claim 3, wherein r equals 1, and the single redundancy chunk for each set of k data chunks contains parity information, and wherein said reconstruction unit (10) is operative to reconstruct a data chunk stored on said memory unit (4) being bypassed using the parity information and other data from the other memory units (4).
Claim 5
Dependent Claim
54 words
A server system as recited in any preceding claim, wherein said control unit (12) comprises means for examining congestion at each of the queues; and means for identifying among said memory units holding data of a given set of data the one of said memory units associated with the queue with the most congestion.
Claim 6
Dependent Claim
65 words
A server system as recited in claim 5, wherein the or each set of data includes a plurality of data chunks and at least one reconstruction chunk; wherein said reconstruction unit (10) is operative to reconstruct the data chunk of at least one memory unit with a congestion greater than that of the memory unit storing the reconstruction chunk by more than a predetermined amount.
Claim 7
Dependent Claim
23 words
A server system as recited in claim 6, wherein congestion is examined by considering at least one of queue length and scheduling priorities.
Claim 8
Dependent Claim
43 words
A server system as recited in any preceding claim, wherein said control unit (12) is operable to examine operational status of said memory units and to decide whether to bypass any of said memory units by using both queue information and operational information.
Claim 9
Independent Claim
93 words
A method of storing data in a number of memory units (4), comprising the steps of: (a) ordering the memory units to form a memory unit sequence; (b) partitioning data to be recorded into a plurality of data chunks; (c) selecting a set of consecutive data chunks of equal size; (d) respectively recording the set of data chunks on a set of the memory units in the sequence; (e) determining a redundancy chunk for the set of data chunks; and (f) recording the redundancy chunk on the next memory unit in the sequence.
Claim 10
Independent Claim
194 words
A method of supplying data streams from memory units (4) to a destination, the data streams being formed from data obtained from the memory units, data being partitioned into blocks, each block being partitioned into a set of data chunks which are stored on a set of different ones of the memory units, at least one redundancy chunk for each set of data chunks being stored in another of the memory units, one chunk per memory unit, the data chunks of each of the blocks of the data along with the at least one redundancy chunk form a group, said method comprising the steps of: (a) maintaining a queue of access requests for each of the memory units (4); (b) examining queue state information of the queues of the memory units associated with a group; (c) selectively reconstructing at least one of the data chunks of the set from the other data chunks of the set and the associated reconstruction chunk based on both queue state information corresponding to the reconstruction chunk and queue state information corresponding to at least one of the data chunks; and (d) supplying the data chunks to the destination.
Claim 11
Dependent Claim
95 words
A method as recited in claim 10, wherein the storing of the data chunks to the memory units is done in accordance with a sequence, and to form the sequence, the method creates a sequence of memory unit numbers which is aperiodic or has a period which is larger than the number of memory units, guarantees a substantially equal number of recurrences of every memory unit number within any subsequence of the sequence having at least some specified length, and ensures at least a certain specified distance between recurrences of the same memory unit number.
Claim 12
Dependent Claim
59 words
A method as recited in claim 10 or 11, wherein said reconstructing step includes the step of determining whether or not to reconstruct at least one of the data chunks by comparing, for the group, the queue state information corresponding to the reconstruction chunk relative to the queue state information corresponding to at least one of the data chunks.
Claim 13
Dependent Claim
15 words
A method as recited in claim 12, wherein the queue state information is queue length.
Claim 14
Dependent Claim
51 words
A method as recited in claim 13, wherein said reconstructing step determines whether or not to reconstruct at least one of the data chunks based on both the queue length corresponding to the reconstruction chunk and the longest queue length among the queue lengths of the data chunks of the group.
Claim 15
Dependent Claim
60 words
A method as recited in claim 10, wherein said reconstructing step comprises dispatching the access requests to said memory units, wherein the priority with which the access requests are dispatched to the corresponding memory unit depends on the time at which a data chunk is required for reconstruction of a redundancy chunk or for said supplying of the data chunk.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.