MR92 Marek, R.; Rahm, E. 1992: Performance Evaluation of Parallel Transaction Processing in Shared Nothing Database Systems. Proc. 4th Int. PARLE Conference, LNCS 605, Springer, 295-310. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
A Simulation Study for Shared-Disk Parallel Database Systems - Kemme (1996) (Correct)
....(million instructions per second) CPU requests are specified by the number of instructions needed to fulfill the request, and served on a FCFS basis. All major steps of the transaction execution require the usage of the CPU. We chose our settings based on former simulation studies [ACL87, Bhi88, MR92, Rah93] and [GR93] At the begin of the transaction BotIns instructions simulate the overhead of the transaction start. The EOT requires EotIns instructions. Each lock request consumes LocLkIns instructions at the LLM, and GlLkIns at the GLM. The overhead for releasing the locks at EOT is ....
R. Marek and E. Rahm. Performance evaluation of parallel transaction processing in shared nothing database systems. In Proc. 4th Int. PARLE Conference, Paris, June 1992. BIBLIOGRAPHY 94
Multi-processor File-System Performance Modelling - McCann, Valsamidis, Bolton, .. (1993) (Correct)
....use of simulation techniques to model the dynamic nature of parallel DBMSs and the problem of measuring low level DBMSs is reduced by employing a simple analytical model. This Hybrid approach has been used successfully before, but on the whole it is rarely used in DBMS performance studies, except [10, 11]. Such modelling techniques, on the other hand, are used extensively in Operating System performance studies, usually involving a trace based model of an existing system [6, 1] Some of these also focus specifically on caching [4, 5, 9, 17] Whilst being very accurate this approach can be overly ....
Marek R., Rahm E., 'Performance Evaluation of Parallel Transaction Processing in Shared Nothing Database Systems'.Proceedings of 4th International PARLE Conference, June 1992
High Performance Transaction Systems on the SB-PRAM - Gemünd, Jakob, Massonne.. (1995) (Correct)
....[32] of the EDS 2 experimental parallel data base system gives an instruction count of 69721, but about 56000 seem to come from message passing and context switching. This is reasonably consistent with our numbers. However, these number differs considerably from numbers published elsewhere. In [17] an instruction count of I = 150000 instructions is stated. This number of instructions is apparently indeed consumed by commercial DB systems, but the reasons for this may lie purely in the long history of these systems. From the literature, we could not even find out what percentage of ....
.... with other systems In table 6, we compare performance data of the SB PRAM with 512 data disks with published data for the following systems: ffl the commercial DBS Oracle implemented on a nCube in [8] ffl EDS in [33, 32] ffl the simulations of the university of Kaiserslautern (KL) in [17] For comparison purposes, we define the quality Q of a system by: Q = p m ; where p is the number of processors and m is the number of RISC MIPS of one processor. Here, we count 1 CISC MIPS as 2 RISC MIPS [18] Thus the CISC processor of the nCube is rated 15 RISC MIPS instead of 7.5 CISC ....
[Article contains additional citation context not shown here]
Marek, R., Rahm, E.: Performance Evaluation of Parallel Transaction Processing in Shared Nothing Database Systems. Lecture Notes in Computer Science, Springer--Verlag (1992).
MPL-Adaptive Algorithms for Multisegmented Disk Caches - Soloviev (1994) (3 citations) (Correct)
....cache algorithms comes from analyses of data placement in parallel database systems. In such systems, data files are usually declustered across many nodes and across several disks within a node ( Livn87] There is a discussion in progress about the merits of full and partial declustering ([Cope88, Mare92, Rahm93, Meht94]) We show in [Solo94] that if the number of concurrent file scans exceeds the number of disk cache segments then full declustering can lead to performance degradation. It is generally accepted that if the number of sequential scans becomes larger than the number of cache segments, then the ....
Marek, R., Rahm, E., "Performance Evaluation of Parallel Transaction Processing in Shared Nothing Database Systems", Proc. 4th Int'l PARLE Conf. (Parallel Architectures and Languages Europe), Springer-Verlag, Lecture Notes in Computer Science 605, 1992.
On the Performance of Parallel Join Processing in Shared.. - Marek, Rahm (1993) Self-citation (Marek Rahm) (Correct)
No context found.
MR92 Marek, R.; Rahm, E. 1992: Performance Evaluation of Parallel Transaction Processing in Shared Nothing Database Systems. Proc. 4th Int. PARLE Conference, LNCS 605, Springer, 295-310.
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