Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1999-11-12
2002-07-16
Choules, Jack (Department: 2177)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000
Reexamination Certificate
active
06421687
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to real-memory data processing systems and, in particular, to distributed data processing systems comprising multi-processing platforms.
BACKGROUND OF THE INVENTION
In the art of data processing, in particular for the processing of vital data such as in telecommunications systems, in banking operations or at the stock market, for example, it is common to use some kind of backup processing and storage in order not to loose vital data in the case of failure of a data processing system or one of its components.
In some applications, dual processing and storage means are used, wherein one of the processing and storage means performs the actual processing of data and another of the processing and storage means operates in a stand-by mode. The stand-by equipment can either operate in an idle mode or parallel to the equipment performing the actual data processing such that, in case of failure, the processing tasks can be taken over by the stand-by equipment without delay or lost of data.
In an other system configuration, multi-processing and storage devices are provided, each performing part of the total data processing. The overall storage capacity of the system is larger than actually required, which provides the opportunity to store backup data distributed over the several storage means of the system. This type of operation is also known as the virtual storage distributed data processing concept, such as disclosed by EP 0 381 644.
In the system of EP 0 381 644 a cluster of processors operates on data structures in virtual storage sharable by each of the plurality of processors. A location table is used in order to store and retrieve data for processing purposes. The system assures the reliability of system-wide shared data structures in the event of failure of one of the processors by maintaining at least two copies of data stored and by maintaining two copies of the location table.
In large virtual storage systems a considerable amount of overhead can be envisaged in order to retrieve the desired data of data structures to be processed, as well as extended access times for storage transactions, increasing the total processing time. Despite the enhanced reliability of the overall data processing system, for high-speed data processing required in modern Intelligent Network (IN) telecommunication processing systems, the virtual storage concept is not always applicable.
WO96/37837 discloses a data base server system having multiple nodes, each comprising its own processing unit, communication equipment for communication with other nodes and storage means. The nodes are divided into at least two independent groups, which share nothing. The storage means of a node each just contain a fraction of the total system data. That is, for N nodes the total system data is divided in N fragments of essentially equal size, called, “primary replica”, and each such fragment is copied and stored as a so-called “standby replica” in the storage means of a node belonging to another group. After failure of a node in a group, fragment replicas which have become unavailable are regenerated and stored on the remaining available nodes in the same group as the failed node.
In this system, because of the spreading of the total system data in fragments over all the nodes, a transaction manager is required for directing data of all queries to the particular processor and storage means comprising the primary replica of a particular data fragment to which a query relates.
Further, in the event of a node failure, the processing load on the node comprising the standby replica is (temporarily) doubled. That is, this node has to process the queries for its primary replica as well as for the standby replica of the primary replica which has become unavailable. This double load situation lasts until the standby replica has been portioned over the remaining available nodes of a group. However, this repair procedure even increases the load on the already double loaded node, which, in general, enhances the risk of failures and decreased processing speed.
Due to the extensive communication required between the nodes of a group and the transaction manager as well as the load doubling of a node in the event of a failure, this data base server system lacks application in modern Intelligent network (IN) telecommunication processing systems, wherein high-speed data processing is a prerequisite.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a method for data processing in a distributed data processing system, comprising a plurality of interconnected processing platforms and storage means, assuring a degree of reliability comparable to the virtual storage concept, while maintaining the processing speed of real-memory systems, although not being cursed with idle backup processing power.
It is a further object of the present invention to provide a method for distributed data processing providing a relatively even load distribution over the processing equipment of the system.
It is another object of the invention to provide a method for distributed data processing providing reliable failure handling in case of failure of a particular processing or storage device or devices.
These and other objects and advantages of the present invention are provided by a method for data processing in a distributed data processing system, comprising a plurality of processing platforms interconnected by a communication network, wherein a platform comprises processor means providing service to a plurality of processes, control means controlling process and system data handling by a platform and storage means allocated to the platform for storing and retrieving system data, the method comprising the step of:
a) storing in the storage means allocated to a platform part of the system data for processing by the platform, and being
characterized by the steps of:
b) duplicating portions of the system data parts stored in storage means allocated to platforms other than the platform of step a);
c) storing the portions of duplicated system data in the storage means allocated to the platform of step a), and
d) processing of a portion of the duplicated system data by the platform of step a) if a platform to which the system data part is allocated corresponding to such portion is not able to process the system data part.
With the method of the invention, part of the total system data is allocated to a processing platform and portions or fragments of such part are duplicated and allocated to other processing platforms of the system. Preferably, each processing platform provides at least a set or a subset of like operations.
In the case of failure of a processing platform data originally allocated to the failing platform of the system is processed by the other platforms where portions of duplicated data reside. That is, the storage means comprising the duplicated or backup data portion of the system data which requires processing. The overall processing is not affected because another processing platform can provide service to portions of duplicated data stored in its storage means under the same conditions as the failing processing platform. Because each processing platform contributes to the overall data processing of the system, no idle processing means are envisaged.
It will be appreciated that, by the portioning of the duplicated system data according to the invention, in the event of platform failure load sharing is achieved because a live processing platform of the system just has to perform part of the processing power of the failing platform.
In an embodiment of the method according to the invention, operating in a system comprising a number of N processing platforms, each of which comprising storage means, N(N−1) different portions of duplicated system data are formed. In order to achieve enhanced reliability the duplicated data are stored such that each platform contains at least N−1 different portions of duplicated system data allocated to other processing pla
Ali Mohammad
Burns Doane Swecker & Mathis L.L.P.
Choules Jack
Telefonaktiebolaget LM Ericsson (publ)
LandOfFree
Data partitioning and duplication in a distributed data... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Data partitioning and duplication in a distributed data..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Data partitioning and duplication in a distributed data... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2843112