Hector Garcia-Molina and Kenneth Salem. "main memory database systems : An overview ". IEEE Transactions on Knowledge and Data Engineering, 4(6), 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....populated tables, which is also disadvantageous. There are also hybrid mappings that try to combine advantage of both approaches [17] In relation based data structures, data compression is often useful for accelerating performance. Conventional databases usually represent data items as strings [7] although it is also possible to use fixed length pointers to compress domain values [16] We have been able to exploit this compact representation of relational data [4] New research in this field shows how to compress the data even further, using advanced compression techniques on the required ....

Hector Garcia-Molina and Kenneth Salem. Main memory database systems: An overview. IEEE Trans. Knowledge Data Eng., 4:509-516, 1992.

....is re ected in an absolute decline in the price of entry level disks. This process has no analogy in semi conductors. The increasing economic viability of main memory databases has prompted a considerable amount of research e ort to be devoted to the investigation of main memory database systems [2], 14] 3 prices.eps Figure 1: Price trends for disk and DRAM store. A typical approach to data representation in such systems is to characterise domain values as a sequence of strings. Tuples are represented as a set of xed length pointers to corresponding domain values [5] Each domain is ....

....backed memory to hold part of a transaction log le, which can be periodically ushed to disk. The important fact is to ensure that there is no disk latency on individual transactions. 2 Implications for Databases Although there has been considerable research into memory resident databases [2][14] most current DBMSs are built around disk resident data. They require external data structures to alleviate the disk I O bottleneck and are optimized for speed rather than disk space occupancy. For instance Oracle Version 6 uses table 4 spaces which are xed sized OS les. Within these, table ....

H. Garcia-Molina and K. Salem. Main memory database systems: an overview. Transactions on Knowledge and Data Engineering, 4(6):509-516, December 1992.

....to port existing databases, such as Oracle) on parallel machines (such as nCube or KSR) Much less effort has been made in finding good architectures for database management systems, which are inherently parallel and thus can take full advantage of parallelism. Main memory resident databases [1, 2, 3] are often considered obsolete, because of their limited capacity of memory and their unability to scale up with growing needs. This objection is not true any more, if a memory resident database is implemented on a parallel architecture, which not only can incorporate substantially large main ....

H. Garcia-Molina, K.Salem: "Main Memory Database Systems: An Overview"; In: IEEE Transactions On Knowledge And Data Engineering, Vol. 4, No. 6, December 1992

....RAM storage has been developed [11] based on logging of database updates and saving complete database images. The system can be used as a single user database or as a multi user server to applications and to other AMOS II systems. The data manager is designed for high performance in main memory [12] and is optimized for efficient execution when the entire database fits in main memory. AMOS II s predecessor, Amos [13] was built on top of the workstation version of the Iris system, WS Iris [14] running on Unix platforms. AMOS II has a completely new kernel developed on a Windows NT 95 ....

....of Copyright # 2001 John Wiley Sons, Ltd. Concurrency Computat. Pract. Exper. 2001; 14:1 21 July 12, 2001 Marked proof Ref: CPE607 23567e Sheet number 8 8 T. RISCH AND V. JOSIFOVSKI function definitions. We use tailored main memory data structure representations of system objects [12], rather than, for example, storing them in relational tables. For example, our object identifiers are represented as variable length records with pointers to data structures representing type information, function definitions, dependent objects, etc. It is crucial that system information is ....

Garcia-Molina H, Salem K. Main memory database systems: An overview. IEEE Transactions on Knowledge and Data Engineering 1992; 4(6):509--516.

....notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and or a fee. ACM SIGMOD 2001 May 21 24, Santa Barbara, California, USA Copyright 2001 ACM 1 58113 332 4 01 05 . 5.00. example, [9, 12, 15, 16]) by providing an approximate order ofmagnitude performance improvement for simple database applications, when compared to disk databases with data fully resident in main memory [2, 29] Adapting main memory database algorithms to become cacheconscious, that is, to perform well on multi level ....

H. Garcia-Molina and K. Salem. Main memory database systems: An overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509--516, December 1992.

.... large and relatively cheap memory makes it possible to have main memory databases (MMDB) where all data reside in main memory, which provides significant additional performance benefits as shown in [8, 16] In fact, MMDB has been receiving the attention of database researchers for the past decade [6, 1, 3, 10, 17, 4]. When the entire database resides in the main memory, related techniques developed under the assumption of disk I O as the main cost of database operations should be reexamined. Among the others, index structures that affect the overall system performance heavily has been one of the research ....

H. Garcia-Molina, K. Salem, "Main Memory Database Systems: An Overview", IEEE Transactions on Knowledge and Data Engineering, Volume 4, Number 6, December 1992, pp. 509-516.

....are partitioned to multiple disks. Hence, some of the AEMs have a disk attached to store passive objects, while some others are used for processing only and they just have local memory and no disk. However, assuming that AEMs have large enough main memory and the appropriate architectural support [14], the same abstract architecture can be used for a main memory database system [8] The existence of multiple AEMs provides a natural declustering for the objects of a class [8] Declustering criteria can be any one of the popular data partitioning strategies of parallel relational database ....

H. Garcia-Mollina and K. Salem, Main memory database systems: An overview, in M.H. Eich, ed., Special section on main-memory databases, IEEE Trans. on Knowledge and Data Engineering 4(6) (Dec. 1992) 509-516.

....access of large objects. To enhance the performance of database systems, apart from traditional optimization techniques, multiprocessor database systems have been proposed, both for relational [4, 7, 8] and object oriented [2, 13, 15] databases. Furthermore, main memory systems have been reported [9, 10] to provide better performance for time lined database applications. Parallel main memory relational database systems have been studied in the context of the PRISMA DB project [1, 17] and also in [5] while the conjunction of sequential object oriented and main memory databases received some ....

H. Garcia-Mollina and K. Salem, Main memory database systems: An overview, in Ref. [9], 509516.

....we consider memory access optimization to be relevant for database performance in general, it is especially important for main memory databases, a field that through time has received fluctuating interest within the database research community. In the 1980s [LC86a, LC86b, Eic89, Wil91, AP92, GMS92] when falling DRAM prices seemed to suggest that most data would soon be memory resident, its popularity diminished in the 1990s, narrowing its field of application to real time systems only. Currently, interest has revived into applications for small and distributed database systems, but also ....

H. Garcia-Molina and K. Salem. Main Memory Database Systems: An Overview. IEEE Trans. on Knowledge and Data Eng., 4(6):509--516, December 1992.

.... real time data is to replace disk with faster main memory for their storage [21] The Main Memory Data Base Management System (MMDBMS) uses main memory as a primary storage for eliminating the cost of disk accesses, which have been known as the main performance bottleneck of disk based DBMS [1] 4][8]. 1 Realtime DBMS Team at ETRI and Data and Knowledge Engineering Lab. at Kangwon National University in Korea have been working together to develop the Tachyon, a next generation MMDBMS. The Tachyon supports a deadline concept because it considers real time applications as its major target. The ....

H. Garcia-Molina and K. Salem, "Main Memory Database Systems: An Overview," IEEE Trans. on Knowledge and Data Engineering , Vol. 4, No. 6, pp. 509-516, 1992.

.... previous note page Date: 1st October 1996 Slide: 15 1 University of Skovde Technology Transfer: Traditional databases Main Memory Databases an overview ffl The purpose of a main memory database (MMDB) is to eliminate disk I O in order to make the database more efficient ffl Impacts on [ Garcia Molina and Salem, 1992 ] purpose of an MMDB because of problems with [ Eich, 1987 ] commit processing recovery mechanism concurrency control storage structure performance API and protection Date: 1st October 1996 Slide: 16 University of Skovde Technology Transfer: Traditional databases 16 ....

....However, there are solutions. It is possible to use incremental recovery if stable main memory is available [ Levy and Silberschatz, 1992 ] The main solution to the second problem is to have the log tail, i.e. the tail of the log (the latest log entries) in stable main memory [ Eich, 1987, Garcia Molina and Salem, 1992]. Moreover, additional hardware for checkpointing and logging may increase the performance. Date: 1st October 1996 Slide: 17 1 University of Skovde Technology Transfer: Traditional databases MMDB Impact on DBMS ffl Concurrency control: granularity of locking entities may be larger than in ....

[Article contains additional citation context not shown here]

Garcia-Molina, H. and Salem, K. (1992). Main memory database systems: An overview. IEEE Trans. on Knowledge and Data Engineering, 4(6):509--516.

....storage requirement grows, require predictable and efficient representation, access methods, and query processing. Few, if any, traditional databases can provide these services in a predictable way [20] A great source of unpredictability, disk accesses, can be avoided by using a main memory DBMS [12]. Active databases have been proposed as an approach to efficient situation monitoring in which the database works as an active component in the system. Most approaches adopt the notion of event condition action rules (ECA) 10] which is used to model reactive behavior. The modeling of reactive ....

H. Garcia-Molina and K. Salem. Main memory database systems: An overview. IEEE Trans. on Knowledge and Data Engineering, 4(6):509--516, December 1992.

....integrity, concurrency and more, Moss, 1985] a book covering nested transactions in depth, Ullman, 1982] another old book written by a well known author. Main memory databases. Articles: Eich, 1987] covering a simulation study of different checkpointing and logging mechansisms; [Garcia Molina and Salem, 1992] providing a good overview of main memory databases; and [Levy and Silberschatz, 1992] providing an interesting mechansism for incremental recovery. Active databases. Articles: Dayal et al. 1988, Chakravarthy et al. 1989] the former is one of the best known articles in this ....

....graphs; Gatziu and Dittrich, 1992, Gatziu and Dittrich, 1993, Gatziu et al. 1994] covering the active database SAMOS. Real time databases: Articles: Ramamritham, 1993a, Ramamritham, 1993b] which give a good overview of the problems in real time databases; Graham, 1992, Kao and Garcia Molina, 1992] additional perspectives on the problems of real time databases. ffl Distributed systems. Books: Mullender, 1994] a very good book covering most of the interesting issues in distributed computing. ffl Event monitoring. Articles: Chakravarthy and Mishra, 1994, Chakravarthy et al. ....

Garcia-Molina, H. and Salem, K. (1992). Main memory database systems: An overview. IEEE Trans. on Knowledge and Data Engineering, 4(6):509--516.

.... the SORAC object oriented database system that we implemented by extending the commercial ONTOS object oriented database [Ont89] with the capability to support object relationships [DPW93] Our new RT SORAC testbed is a main memory database (for a survey of main memory database techniques, see [GMS92]) which follows our object oriented real time database model of Section 2. The RT SORAC system is implemented on the Lynx real time operating system [GL91] which is compliant with the proposed POSIX 1003.4a real time operating system standard [IEE90] We have extended Lynx s POSIX 1003.4 ....

Hector Garcia-Molina and Kenneth Salem. Main memory database systems: An overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509--516, December 1992.

....for real world applications. Our performance study shows that DSFC speeds up the restart operation with minimal overhead. Index Terms: Main memory databases, recovery, fuzzy checkpointing, logging. 1 Introduction Main Memory Databases (MMDB) have gained much attention for more than a decade [5, 7, 8, 9, 14, 15, 16, 17, 19], with several MMDB products [2, 27] With all or a major portion of the database residing in main memory, MMDBs can achieve very high performance. Due to the volatility of main memory, however, the recovery issues for MMDBs are especially crucial. The recovery scheme for an MMDB should a ect ....

H. Garcia-Molina and K. Salem. Main memory database systems: An overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509-516, December 1992.

....IIS 98 12014, and by NSF CISE award CDA 9625374. 0 1 Introduction As random access memory gets cheaper, it becomes increasingly a#ordable to build computers with large main memories. There are many existing applications with large datasets of the order of several gigabytes which can fit in RAM [GMS92] Recently, the Asilomar Report ( BBC 98] predicts Within ten years, it will be common to have a terabyte of main memory serving as a bu#er pool for a hundred terabyte database. All but the largest database tables will be resident in main memory. An important factor in main memory data ....

Hector Garcia-Molina and Kenneth Salem. Main memory database systems: An overview. IEEE Transactions on knowledge and data engineering, 4(6):509--516, 1992.

....ASC 9612099, and MIP 9619351, DARPA Contract DABT63 95C 0097, NASA Contract NAG 1 613, and gifts from Intel and IBM. Pedro Trancoso was also supported by the Portuguese government under scholarship JNICT PRAXIS XXI BD 5877 95. 2 Now at Intercollege Limassol, Cyprus. by Garcia Molina et al. [3] suggested that the disk resident assumption inhibits higher performance. It is necessary to redesign the database to fully exploit main memory residence. The resulting systems are called Memory Resident Database (MRDB) systems. MRDB systems have been studied by several researchers [5, 21] and ....

....of its plans is a hard coded condition. This condition asserts that a Hash Join algorithm can be used only if the inner input table is smaller than the outer input table. This DRDB condition, which tries to avoid costly data spills to disk, ends up disabling good plans. Garcia Molina and Salem [3] also report this problem in other DRDB systems. Overall, we conclude that the plans generated for a memory resident system deliver better performance. Figure 6 (b) shows that, among memory resident plans, the m plan is always the worst. Clearly, therefore, optimizing for the minimum number of ....

H. Garcia-Molina and K. Salem. Main Memory Database Systems: An Overview. IEEE Transactions on Knowledge and Data Engineering, pages 509-516, December 1992.

....Keywords Distributed database system, Fault tolerance, Group commit, Main memory database, Precommit. 1 Introduction The memory resident database or main memory database (MMDB) has a potential for significant performance improvement over conventional database where the data reside in the disk[3, 6, 2]. Since there is virtually no I O overhead for reading and writing the database, the MMDB shows 1 faster response time and improved throughput[10, 12, 17, 9] With the falling price of RAM and the rising demand of high performance system, MMDB has become a real alternative for database systems. ....

H. Garcia-Molina and K. Salem. Main memory database systems: An overview. Trans. on Knowledge and Data Engineering, 4(6):509--516, December 1992. 17

....the same trend, the price of memory is falling at roughly the same rate as the price of disks. This growth rate in amount of memory per dollar cost has led many to proclaim the advent of main memory databases to take over many of the functions that traditionally required the use of disk storage [Garcia Molina92, Jagadish94]. This trend has clearly helped transaction process ing workloads with relatively small working sets proportional to the number of live customer accounts or the number of unique items in a store s inventory, for example but does not address the data growth when historical records all ....

Garcia-Molina, H. and Salem, K. "Main Memory Database Systems: An Overview" TKDE 4 (6), December 1992.

....optimal path to their destinations in terms of travel distance, travel time and other criteria. Due to the dynamic nature of the ITS database and the frequency and nature of queries, it is necessary to explore alternatives to traditional disk resident databases (DRDB) Main memory databases (MMDB) [4] store most of their content permanently in main memory, resulting in high speed access to that data. Data stored in memory is hot if it is frequently accessed, while data kept on disk is cold if it is accessed less frequently. Main memory databases can store both types of data. MMDBs differ from ....

Hector Garcia-Molina and Kenneth Salem. Main Memory Database Systems: An Overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509--516, 1992.

....based on CPU cost optimization for query processing have been suggested [26, 83, 56] However, modeling CPU costs is not an always easy task. Costs may vary substantially depending on the hardware platform, the style of programs that carry out the operations, and the overall software design [30]. In addition, there are interesting trade offs between the amount of CPU processing required and the memory buffer space reserved for indexing purposes. In conventional query optimization, there have been numerous efforts to efficiently process queries and in particular joins by ....

....may not be a reasonable option as it could impose additional and unnecessary overheads in terms of CPU processing and space used. Instead, the outer relation can be traversed sequentially and the joining attribute value can be used to access the appropriate joining tuples from the inner relation [30]. This access is facilitated by the traversal of navigational pointers provided by the MMDs, as mentioned earlier in the context of Dal i and Starburst. Hence, the sort merge approach is not used for join processing in main memory databases as it not only requires extra space 8 to accommodate ....

[Article contains additional citation context not shown here]

H. Garcia-Molina and K. Salem. Main Memory Database Systems: An Overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509--516, December 1992.

....large and relatively cheap memory makes it feasible to store database entirely or almost entirely in memory. With data directly accessed from memory, main memory database can provide better performance for real time applications. There have been many designs and implementations of main memory DBMS[2]. These include IMS VS Fast Path[3] TPK[4] OBE[5] System M[6] and main memory storage systems, such as the Starburst s memory resident storage component[7] and Dali[8] These systems, however, were designed to maximize transaction throughput rather than to meet the timing constraints of ....

H. Garcia-Molina and K. Salem, "Main Memory Database Systems: An Overview," IEEE Trasactions on Knowledge and Data Engineering, vol. 4, Dec. 1992.

No context found.

Hector Garcia-Molina and Kenneth Salem. "main memory database systems : An overview ". IEEE Transactions on Knowledge and Data Engineering, 4(6), 1993.

No context found.

H. Garcia-Molina and K. Salem. Main Memory Database Systems: An Overview. IEEE Transactions on Knowledge and Data Engineering, 4(6):509--516, 1992.

No context found.

Garcia-Molina, Hector and Salem, Kenneth, "Main Memory Database Systems: An Overview", Transactions on Knowledge and Data Engineering, Vol. 4, No. 6, December 1992.

First 50 documents  Next 50

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