Intel X25-m Mainstream SATA SSD
Published by Mathias Seiler on 08.09.08iozone ops/s
This test shows the same characteristic like the one before. The difference is that we get the number of opertations per second.
| iozone iops | Intel SSD | Seagate Cheetah |
| sequential write 4 KB | 18'808 | 20'351 |
| sequential write 16 KB | 4'735 | 5'185 |
| sequential write 64 KB | 1'209 | 1'289 |
| sequential write 96 KB | 792 | 838 |
| sequential write 128 KB | 587 | 602 |
| sequential read 4 KB | 58'241 | 15'792 |
| sequential read 16 KB | 14'778 | 4'146 |
| sequential read 64 KB | 3'663 | 1'019 |
| sequential read 96 KB | 2'428 | 677 |
| sequential read 128 KB | 1'844 | 483 |
| random write 4 KB | 6'297 | 609 |
| random write 16 KB | 4'574 | 560 |
| random write 64 KB | 1'229 | 460 |
| random write 96 KB | 823 | 372 |
| random write 128 KB | 616 | 325 |
| random read 4 KB | 5'349 | 886 |
| radnom read 16 KB | 4'325 | 843 |
| random read 64 KB | 2'501 | 690 |
| random read 96 KB | 1'978 | 571 |
| random read 128 KB | 1'614 | 534 |
The bigger the data blocks the lower the resulting value is logical because the throughput is constant and get much higher priority. Very astonishing is the number of operations per second with small blocksizes! So many operations per second can't be cached efficiently so the cache influence is near zero because the "Cache Hit Rate" sinks.
Especially the value at sequential reading can convince (~58 kOps/s). To compare: A Seagate Cheetah® 15K rpm SAS does sequentially reading a maximum of ~15 kOps/s in an Apple XServe.
Only in sequential writing the Seagate drive sees land against the Intel SSD. All the other tests speak a very clear language.
Also the throughput looks only good when the server hard drive from Seagate is writing.
Discuss this article in the forum
| Navigate through the articles | |
Western Digital Raptor 150GB |
Seagate Barracuda 7200.11 1.5 TByte
 |
|
|