This is a temporary note to point you to a request from James Morle for a bit of community effort. As spinning rust is gradually replaced by semi-conductor devices, transport latency becomes more significant, so James is trying to get a handle on this aspect of I/O. If you want to help build a coherent picture, read this article and see what you can do about running the test.
Judging from the click-through count that WordPress is reporting, there’s not much interest in James’ project. Since the only way to get good information on a topic like this is to have a good sample set, it’s in everyone’s interest to help if they can. The only alternative to a scientific approach based on collectint data is to wait for someone to feed you bullshit based on marketing material.
Comment by Jonathan Lewis — February 27, 2012 @ 7:18 pm UTC Feb 27,2012 |
Jonathan,
Looks like James’s test is a bit flawed, and it won’t measure what you might expect. Have you seen the comments by Kevin on his blog?
Comment by Christo Kutrovsky — February 29, 2012 @ 1:34 pm UTC Feb 29,2012 |