Here you can see the minimum and maximum transfer performance of all configurations in one single chart. The four drive, four channel setup is indeed almost four times faster than the single DTLA.
Disk Access Time
It was predictible that the access time would not be quicker in a stripeset. Though the data throughput is considerably higher, the drives cannot possibly position their heads any faster. This not only applies to software RAID, but also for all kinds of hardware solutions. Expensive adapters may have RISC controllers with a lot of cache memory and sophisticated algorithms, but they cannot solve the basic problem. Conclusion: If you are running demanding applications that frequently access data bases (e.g. a webserver), try to get drives that have quick access times.
- IDE RAID Without Additional Hardware: Do It Yourself With Windows 2000
- Data Fragmentation: A Performance Killer
- IDE Vs. SCSI RAID
- Software RAID Requirements
- Perfidies Of The Software RAID
- Test Setup
- Benchmark Results
- Data Transfer Diagrams, Continued
- Data Transfer Comparison Chart
- Application Benchmark: WinBench 99 Disk Winmarks
- SiSoft Sandra 2001 Performance Index