摘要:
A deadlock detection system each comprise a lock control table having entries corresponding to a resource when a plurality of resource requesters share a plurality of resources, a resource requester identifier list connected to a table of the lock control table, and a lock controller for managing an exclusive control by using the lock control table and the resource requester identifier list and for detecting a deadlock by examining a sequence inversion among resource requester identifiers on a list among competing resource requesters. An element of the resource requester identifier list can distinguish a shared lock from an exclusive lock by transferring a lock list obtained through a breakdown according to the content of a shared lock, and by independently examining the sequence inversion of resource requester identifier of the list among competing resource requesters.
摘要:
A parallel sorting system is disclosed, having N (=2.sup.n) processing devices, each of which comprises a sort element storing unit, an inner sorting unit, a transposing unit, and a comparing and substituting unit. In the sort element storing unit, sort elements/N are stored. Sorting process is performed for n stages. In each stage, the inner sorting unit sorts the sort elements, and the transposing unit exchanges sorted elements at a transposing position with another processing device through a network. Thereafter, the comparing and substituting unit compares the two sets of sorts elements. This process sequence is repeated for all stages. And, after the transposing process and the inner sorting process are performed, the sorting process are completed. In this system, the number of communication times is reduced to 2n.
摘要:
A database system including a transaction database system processing device for executing a transaction database process and a decision support database system processing device for executing a decision support process. The database system further includes a transaction database storage device, provided with the transaction database device, for retaining a transaction database created and updated by the transaction database process, a first and second database storage device, each provided with the decision support database device, and each for retaining a first and a second decision database, respectively, for use in a decision support process, a delayed updating device, provided with the decision support database device, for receiving a log created based on the transaction database at each of a predetermined time interval, and for alternatively supplying both the log received at a current predetermined time interval and the log received immediately preceding the current predetermined time interval to the first database storage device and to the second database storage device at each predetermined time interval, and a retrieving device, provided with decision support database device, for executing a retrieving process for the second decision database stored in the second database storage device when the delayed updating device supplies both logs to the first database storage device and for executing a retrieving process for the first decision database stored in the first database storage device when the delayed updating device supplies both logs to the second database storage device.
摘要:
A plurality of unit streams which can be simultaneously supplied at a reference speed are determined and a schedule forming section previously forms a schedule for allocating time-dependent uses of a memory unit, an input/output unit, and a buffer memory to all of unit streams. When a stream is actually requested, un-used unit streams of the necessary number are allocated in accordance with a request speed by a stream allocating section. The request stream is supplied in accordance with the schedule of the allocated unit streams.
摘要:
A storage system having a plurality of storages. The each of the storages include a memory and a processor coupled to the memory. The processor executes a process including transmitting an update request for data which is commonly stored in the plurality of storages according to a predetermined transmission order indicating a path to transfer the update request. The process includes updating data when receiving an update request from another storage. The process includes changing the predetermined transmission order to a transmission order in which one or more storages included in the path are excluded according to the number of times the update request for the data is received.
摘要:
A number-of-slices collector collects number-of-slices information indicating the number of free slices, which are not assigned to a storage areas of a logical disk, from each of a plural disk nodes, divides the storage area of the corresponding storage device in units of slice, and stores an assignment relation of the slices with respect to the storage areas of the logical disk. A disk node selector selects a source disk node from among the disk nodes having the free slices, and requests a slice identifier to identify the free slice for the selected source disk node. The source disk node serves as a source providing the slice to be assigned. A slice assigner receives the slice identifier from the source disk node, and determines an assignment relation of the free slice, which is denoted by the received slice identifier, with respect to the storage areas of the logical disk.
摘要:
A computer executes a data allocation control program to control allocation of data in a plurality of disk nodes. A redundancy restoration module executes a redundancy restoration procedure by commanding disk nodes to create a new copy of redundancy-lost data. An error message reception module receives a write error message indicating a write error in a storage device during the redundancy restoration procedure and records an identifier of the faulty storage device in an error record memory. A copy command module identifies accessed data in the faulty storage device and commands a relevant disk node to create a copy of the identified data in the faulty storage device in the case where there is no redundant copy of the data.
摘要:
In a computer executing processes based on a storage management program, an attribute collection unit collects attributes of storage nodes. A group creation unit creates at least two groups that include the storage nodes, in accordance with the attributes of the storage nodes collected by the attribute collection unit. A data allocation unit allocates distributed data and redundant distributed data to the groups created by the group creation unit in such a manner that the distributed data and the redundant distributed data having the same content as the distributed data are placed in different groups.
摘要:
A non-transitory, computer readable storage medium storing a program for a computer being capable of accessing a group of nodes having divided-data structures obtained by dividing a data structure having a transposed hierarchical bloom filter of a hierarchical bloom filter and sets of data blocks by a predetermined bit width, wherein the hierarchical bloom filter has h-stage bloom filter arrays each having a bit width of “s”, the hth-stage bloom filter array having bits indicating false positives or negatives and the pth-stage bloom filter array having n=d[h-(p-1)] bloom filters (where 1≦p≦h) each having a bit width of m=s/d[h-(p-1)] (where d indicates a division number for the hth-stage bloom filter array), the transposed hierarchical bloom filter is obtained in the bloom filters in the bloom filter array, and the set of data blocks corresponds to the first-stage bloom filter array of the hierarchical bloom filter.
摘要翻译:存储用于计算机的程序的非暂时的计算机可读存储介质,其能够访问具有分割数据结构的一组节点,所述分割数据结构通过划分具有分层布隆过滤器的转置的分层布隆过滤器的数据结构和数据块集合 其中分层布隆过滤器具有每个具有位宽s的h级布隆过滤器阵列,第h级布隆过滤器阵列具有指示误报或否定的位,并且具有n位的第p级布隆过滤器阵列 = d [h-(p-1)]绽放滤波器(其中1≦̸ p≦̸ h)具有位宽m = s / d [h-(p-1)](其中d表示 第一阶段的布隆过滤器阵列),在布隆过滤器阵列的布隆过滤器中获得转置的分层布隆过滤器,并且数据块集合对应于分层布隆过滤器的第一阶段布隆过滤器阵列。
摘要:
Processing time required from transmission of a request packet including a processing request from a client, until return of a reply packet in which internal information including an operation status of a plurality of servers is added to an execution result of processing corresponding to the processing request, obtained by the servers in cooperation with each other, is calculated by a function incorporated beforehand in the respective servers. Moreover, the internal information added to the reply packet is extracted and stored in an internal information DB in association with the calculated required time of the processing. Furthermore predetermined statistical processing is performed with respect to the internal information and the required time associated with the internal information stored in the internal information DB to analyze the operation status of the plurality of servers.