The database test is a completely random set of operations, of which 67% are reads. It works with 8 KB block sizes, on which NTFS can capitalize on the non-compressed Samsung architecture, while the SandForce-based drive performs similarly on NTFS and eFAT. FAT32 is barely worth noting here.
The Web server workload does not execute writes, so it delivers similar performance across all the file systems.
The workstation workload patterns split read and write operations 80/20% respectively, with random and sequential operations also split 80/20%. It involves block sizes of 64, 128, and 256 KB, which is why FAT32 shows performance limits again, since it does not support blocks that large.
- What's A File System? Does It Matter?
- File Systems: FAT32, NTFS, exFAT, and HFS+
- Test SSDs: Samsung 830 And Zalman F1 Series
- Test Setup And Benchmarks
- AS SSD: Random And Sequential Throughput
- AS SSD: Access Time, Copy Benchmark, And Overall Score
- CrystalDiskMark: Random And Sequential Throughput
- Benchmark Results: Iometer 4 KB Random And Streaming Read/Write
- Benchmark Results: Iometer Workload Tests
- Benchmark Results: PCMark 7
- Stick To NTFS On Windows