Oracle Log Buffer Queueing
The purpose of this document is to investigate Oracle database log buffer queuing and its affect on the ability to load data using a specialized data loading system. Experiments were carried out on a Linux system using an Oracle 9.2 database. Previous experiments on a Sun 4800 running Solaris had shown that 100,000 entities per minute was an achievable rate. The question was then asked, can we do this on Linux, and where are the bottlenecks? A secondary question was also lurking, how can the loading be further scaled to handle even higher throughput requirements? Testing was conducted using a Dell PowerEdge 6650 server with four CPUs and a Dell PowerVault 220s RAID array with 14 36GB drives and 128 MB of cache. Oracle Enterprise Edition 9.2.0.4 was used for the database and Red Hat Linux Advanced Server 2.1 was used for the operating system. This document will detail the maximum observed throughputs using the same test suite that was used for the Sun tests. A detailed description of the testing performed along with an analysis of bottlenecks encountered will be made. Issues related to Oracle and Linux will also be detailed and some recommendations based on the findings.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- US Department of Energy (US)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 15015171
- Report Number(s):
- UCRL-TR-209462; TRN: US200509%%363
- Resource Relation:
- Other Information: PBD: 8 Dec 2004
- Country of Publication:
- United States
- Language:
- English
Similar Records
Oracle and the end-user: Practical implementation of applications in a multi-user, multi-platform environment
Distributed data access in the sequential access model at the D0 experiment at Fermilab