Weikum, P. Zabback, and P. Scheuermann. "Dynamic file allocation in disk arrays". Proceedings of the 1991 ACM SIGMOD International Conference on Management of Data, pp. 406-415, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....ratio in the hot area than in the cold area, which leads to low garbage collection costs in the hot area. 4.2 The method of clustering hot blocks We have to be able to distinguish between hot blocks and cold blocks. If the access locality is considered static or predictable, we can use heat [12], which is the measure of the number of accesses over a certain period. However, hot blocks may change. In this case, it is difficult to determine an effective method for heat estimation. As described above, almost all blocks written to by normal write accesses can be thought as hot if there are ....

G. Weikum, P. Zabback, and P. Scheuermann. Dynamic File Allocation in Disk Arrays. In Proc. of ACM SIGMOD, pp. 406--415, May 1991.

....4.1 where we show that Ergastulum is faster and generates as good or better designs. Existing solutions to the file assignment problem [11, 35] use heuristic optimization models to assign files to disks to get improvements in I O response times. The file allocation schemes described 23 in [12, 24] automatically determine an optimal stripe width for files, and stripe those files over a set of homogeneous disks. They then balance the load on those files based on a form of hotspot analysis, and swapping file blocks between hot and cold disks. Ergastulum can select the appropriate number ....

P. Zabback G. Weikum and P. Scheuermann. Dynamic file allocation in disk arrays. In SIGMOD Conference, pages 406--415, 1991.

....on an economic model that factors in the tradeoffs between the service quality and the cost. Existing solutions to the file assignment problem [13, 34] use heuristic optimization models to assign files to disks to get improvements in I O response times. The file allocation schemes described in [16, 28] will automatically determine an optimal stripe width for files, and stripe those files over a set of homogeneous disks. They then balance the load on those files based on a form of hotspot analysis, and swapping file blocks between hot and cold disks. Hippodrome can expand or contract the ....

P. Z. G. Weikum and P. Scheuermann. Dynamic file allocation in disk arrays. In SIGMOD Conference, pages 406--415, 1991.

.... and constraint equations expressions be used to solve the self managing storage system problem, i.e. some workloads units have been assigned and new ones are coming in; how should it be determined if the assigned workload units should be moved around The ideas for dynamic data placement from [Weikum90] might be helpful. Wolf89] uses a variable which they call tweak limit that presents the number of files which can be moved from their current disks when their system is changing assignments as a result of a changing workload; we might want to do something similar. 9 Conclusions This document ....

Gerhard Weikum, Peter Zabback, and Peter Scheuermann. Dynamic file allocation in disk arrays. Technical report 147. Department of Computer Science, ETH Zurich, CH8092 Zurich, Switzerland, December 1990. 36

....and size constraints. Lastly, these techniques are not designed for schema partitioning. Index specific load balancing based declustering methods have been proposed for B tree [30] Rtree [20] and the temporal index [25] etc. Dynamic file allocation methods are proposed in the FIVE system [34]. These methods are incremental in nature to balance the load (e.g. storage, I O time) in various partitions for a local window (i.e. a subset of existing data items in partitions) around the new data item. The incremental nature of the load balancing methods allows them to work well with indexing ....

G. Weikum, P. Zabback, and P. Scheuermann. "Dynamic File Allocation in Disk Arrays". In Proc. of Intl Conference on Management of Data. ACM SIGMOD, 1991.

....and size constraints. Lastly, these techniques are not designed for schema partitioning. Index specific load balancing based declustering methods have been proposed for B tree [31] Rtree [20] and the temporal index [25] etc. Dynamic file allocation methods are proposed in the FIVE system [33]. These methods are incremental in nature to balance the load (e.g. storage, I O time) in various partitions for a local window (i.e. a subset of existing data items in partitions) around the new data item. The incremental nature of the load balancing methods allows them to work well with indexing ....

G. Weikum, P. Zabback, and P. Scheuermann. "Dynamic File Allocation in Disk Arrays". In Proc. of Intl Conference on Management of Data. ACM SIGMOD, 1991.

....schemes are based on the heat of the file and the cassette tape. The original concept of heat and temperature was introduced for file management on disk arrays by G. Copeland et al.[10] G. Weikum et al. extended the research to dynamic relocation of data on disk arrays when the file is expanded[11]. As far as the authors know, heat based file relocation research has been confined to secondary storage devices. Because of the many differences between disk arrays and tape archivers, which we will discuss in this paper, we have derived our own heat based schemes for tape archivers. There have ....

G. Weikum, P. Zabback, and P Scheuermann. "Dynamic file allocation in disk arrays". In Proceedings of the 1991 ACMSIGMOD International Conference on Management of Data, pages 406--415, Denver, Colorado, May 1991.

....does not imply that the real time bandwidth will be achieved especially when the storage device bandwidth is much smaller than the required object retrieval rate as is pervasive in multimedia environments. Several file system level approaches ( e.g. Anderson Osawa, 1992; Rangan Vin, 1991; Weikum, et al. 1991]) have been utilized for efficient management of data for local and remote users. For example, Weikum, et al. 1991] deals with the dynamic allocation in disk arrays for complex objects but does not address the diverse characteristics of multimedia objects and associated problems with storage ....

....smaller than the required object retrieval rate as is pervasive in multimedia environments. Several file system level approaches ( e.g. Anderson Osawa, 1992; Rangan Vin, 1991; Weikum, et al. 1991] have been utilized for efficient management of data for local and remote users. For example, [Weikum, et al. 1991] deals with the dynamic allocation in disk arrays for complex objects but does not address the diverse characteristics of multimedia objects and associated problems with storage device configurations. In [Anderson Osawa, 1992; Rangan Vin, 1991] the authors have proposed some storage ....

Weikum, G., Zabback, P., and Scheuermann, P. (1991). "Dynamic File Allocation in Disk Arrays," in Proc. 1991 ACM SIGMOD Intern. Conf. om Management of Data, Denver, Colorado, pp.

....than a disk page. 10 Frequency of access. techniques and evaluate them with metrics such as: working set size [Den68] and hit ratio. 7.2.2 Assigning Objects Disk Pages to Disks Assigning objects disk pages to multiple disks have been studied before. Copeland et al. [CABK88] and Weikum et al. [WZS93] introduced techniques to assign objects and disk pages to disks aiming to balance the load. These techniques were based on the heat of objects. However, these techniques considers neither the frequency of traversals nor the hiccup free display constraint. Load balancing techniques do not ....

G. Weikum, P. Zabback, and P. Scheuermann. Dynamic file allocation in disk arrays. In Proceedings of the 1993 Foundations of Data Organization and Algorithms (FODO) Conference, October 1993.

....spanning paths were proposed in [8] to divide a set of given data items into two similar groups. Index specific load balancing based declustering methods have been proposed for B tree [22] Rtree [13] and temporal index [17] etc. Dynamic file allocation methods are proposed in FIVE system [24]. These methods are incremental in nature to balance the load (e.g. storage, I O time) in various partitions for a window (i.e. a subset of existing data items in partitions) around the new data item. The incremental nature of the load balance method allows them to work well with indexing methods ....

G. Weikum, P. Zabback, and P. Scheuermann. "Dynamic File Allocation in Disk Arrays". In Proc. of Intl Conference on Management of Data. ACM SIGMOD, 1991.

....degree of allocation. It is important to maintain the degree of allocation of relations because it has a direct bearing on the response time of transactions. The methods in their paper can unintentionally cause the degree of allocation of relations to change. This is not a desirable feature. In [19], the authors present a method to determine the degree of allocation of a relation. Their method assumes the system is operated in single user mode, and all read and write requests are for the entire file, rather than for fractions of files as in a database environment. The authors also propose ....

....at the disks which hold the relation fragments. Table 3 shows the number of items requested by each query type from various relations. The following method is used to determine the degree of allocation for each relation in the TPC C benchmark. Earlier methods to determine the degree of allocation [19] were geared towards a single user system and for one relation at a time. This method is geared towards multiple user system and can determine the degree of allocation for all the relations in the database schema at the database design time. It is designed to minimize the response times for the ....

G. Weikum, P. Zabback, and P. Scheuermann. Dynamic file allocation in disk arrays. In Proceeding of the ACM SIGMOD Conference, 1991.

....placement of data is dependent on these dynamic factors. Consequently, there is no one placement of relations that is optimal for the lifetime of the database system. In order to sustain the performance of the system, data has to be frequently moved across disks in order to load balance the disks [19, 20]. It is common that relations have many indexes on them. For example, Sybase SQL server allows upto 250 indexes on a relation [3] Whenever data is moved across nodes, the indexes on the relations have to be modified too. For example, if 10000 records are to be moved, and there are 10 indexes on ....

....conducted the experiments. In section 6, we make a performance evaluation of BULK and OAT. In section 7, we make a qualitative comparison of BULK and OAT, and state our conclusions in section 8. 2 Related Work Though there has been much work in the area of on line reorganization in recent years [20, 9, 16, 6, 11, 12, 7, 8], there has hardly been any work that considers index modification overhead during on line reorganization. Perhaps [11] is the only paper that discusses a solution for on line index modification. However, the techniques in the paper are limited to centralized databases and requires the use of ....

G. Weikum, P. Zabback, and P. Scheuermann. Dynamic file allocation in disk arrays. In ACM SIGMOD Proceedings, pages 406--415, 1991.

....by using data partitioning and function partitioning [5, 18] In order to achieve high performance, the data partitioning becomes more crucial than the function partitioning because of the data intensive characteristic of database application workloads. Thus, several data partitioning strategies [3, 6, 7, 8, 9, 11, 19] have been studied in the context of parallel database systems. In a database system (especially in an OLTP environment) the index file is a very important data structure which significantly affects system performance. In spite of its importance, the problem of how to partition an index file in a ....

....horizontal declustering for exploiting all the I O bandwidth provided by the hardware and employs a new availability technique termed chained declustering [9] Several performance studies [6, 7] have been done in order to show the performance trade offs of the different declustering strategies. In [19], Weikum et al. studied the problem of dynamic file allocation in disk arrays. Their work is different from the others in the fact that they consider dynamic file allocations while the previous work assumes static file allocations. When we allocate a file dynamically, there might not be enough ....

Gerhard Weikum, Peter Zabback, and Peter Scheuermann. Dynamic File Allocation in Disk Arrays. In Proceedings of ACM SIGMOD- International Conference on Management of Data, pages 406--415, 1991.

....or the joins involve only one attribute. A large number of methods have been proposed in the past, aiming to achieve good declustering for partial match queries, that refer to several attributes. Almost all these methods focus on Cartesian Product Files [8] Among the few exceptions is the work in [22], where records are dynamically relocated, to avoid hot spots . All the rest of the declustering methods assume that the allocation of buckets to disks does not change over time. A survey of such declustering methods can be found in [9] There, the methods are roughly grouped into the following ....

Gerhard Weikum, Peter Zabback, and Peter Scheuermann. Dynamic file allocation in disk arrays. Proc. ACM SIGMOD, pages 406--415, May 1991.

....the automatic loop in section 5.2.3 in its management of compute and network resources. Existing solutions to the file assignment problem [DF82, Wol89] use heuristic optimization models to assign files to disks to get improvements in I O response times. The work described on file allocation in [GWS91, SWZ98] will automatically determine an optimal stripe width for files, and stripe those files over a set of homogeneous disks. They then balance the load on those files based on a form of hotspot analysis, and swapping file blocks between hot and cold disks. Hippodrome can expand or ....

Peter Zabback Gerhard Weikum and Peter Scheuermann. Dynamic File Allocation in Disk Arrays. In Proceedings of the 1991 SIGMOD Conference, pages 406--415, 1991.

....the various blocks or extents is often distributed in a highly non uniform manner. 50 1.3 Contribution and outline of the paper This paper addresses several key issues in the automatic tuning of parallel storage systems. Various aspects of earlier versions of our approach have been published in [71,72,79,80], and [78,84] give an account of the COMFORT project where this work has been applied. In this paper, we describe automatic tuning methods that are more advanced than the those used earlier and we emphasize their interaction in an actual system. We provide guidelines for potential system ....

....model determines a degree of parallelism that attempts to minimize the multi user response time; given this degree of parallelism, the striping unit and striping width easily can be tuned. The model and the derived data partitioning method are significant extensions to the method outlined in [79]. In particular, our new algorithm explicitly takes into account queueing delays by providing a computationally tractable analytical approximation for the underlying fork join queueing model [62] In contrast to the modeling approach of [48] that uses a closed queueing model, our method is based ....

[Article contains additional citation context not shown here]

Weikum G, Zabback P, Scheuermann P (1990) Dynamic File Allocation in Disk Arrays. Proceedings of the SIGMOD International Conference on Management of Data, pp 406--415; extended version available as: Technical Report No. 147, Computer Science Dept., ETH Z urich

....of servers across which the file is spread. Our approach draws on the option of redistributing data by means of bucket migration. Data migration as a means for load balancing of data management systems has been considered in a number of papers on multi disk and distributed file systems (e.g. [WSZ91, WJ92, HW94, SWZ94]) However, the underlying file model of these approaches is that of Unix files (i.e. bytestrings) whereas we assume record structured files with key based access. To the best of our knowledge, none of the previous approaches to distributed record structured file organizations has considered the ....

G. Weikum, P. Scheuermann, P. Zabback, Dynamic File Allocation in Disk Arrays, ACM SIGMOD Conference, Denver, 1991.

....itself is not sufficient for this goal. Additional methods for load balancing are called for, regardless of whether data partitioning is used or not. 3 Load Balancing The load balancing component of our intelligent file system consists of two independent modules: one that performs file allocation [19] and the second that performs dynamic redistribution of data [17] After the decision has been made to decluster a file over a particular number of disks, all striping units of the file that are to be mapped to the same disk are combined into a single allocation unit called an extent. The file ....

....extents are allocated in sort order. For each extent allocation, the algorithm selects the disk with the lowest accumulated heat among the disks which have not yet been assigned another extent of the same file. We have extended this greedy algorithm in order to deal with dynamic file allocation [19]. Since our algorithm makes no assumptions about the heat of a new file at allocation time, the sorting step is eliminated and the algorithm only uses the information about the heat of the files which have been allocated already and for which statistics are available. The disk selection can be ....

Weikum, G., Zabback, P., and Scheuermann, P., "Dynamic File Allocation in Disk Arrays," Proc. ACM SIGMOD Int. Conf., 1991.

No context found.

Weikum, P. Zabback, and P. Scheuermann. "Dynamic file allocation in disk arrays". Proceedings of the 1991 ACM SIGMOD International Conference on Management of Data, pp. 406-415, 1991.

No context found.

Gerhard Weikum, Peter Zabback, and Peter Scheuermann. Dynamic file allocation in disk arrays. In Proc. ACM SIGMOD, pages 406--415, 1991.

No context found.

Gerhard Weikum, Peter Zabback, and Peter Scheuermann. Dynamic file allocation in disk arrays. In Proc. ACM SIGMOD, pages 406--415, 1991.

No context found.

G. Weikum,P. Zabback, and PScheuermann. "Dynamic file allocation in disk arrays". In Proceedings of the

No context found.

G. Weikum, P. Zabbak, and P. Scheuermann. Dynamic file allocation in disk arrays. Procs. of the ACM SIGMOD Conference, 1991.

No context found.

G. Weikum, P. Zabbak, and P. Scheuermann. Dynamic file allocation in disk arrays. In Proceeding of the ACM SIGMOD Conference, 1991.

No context found.

Gerhard Weikum, Peter Zabback, and Peter Scheuermann, "Dynamic File Allocation in Disk Arrays", Proc. of the ACM SIGMOD Conf., pp. 406-415, ACM Press, Denver, May 1991.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC