<%BANNER[top_768x90]%>
<%BANNER[banner_468x60_h]%>
<%BANNER[article]%>

Articles: Storage

<%BANNER[fp_160x600_r_1]%>
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 ]

Multi-Threaded Read & Write Patterns

The multi-threaded tests simulate a situation when there are several (from one to four) clients accessing the drive at the same time and the drives are processing requests for sequential reading or writing.

So, I use IOMeter to send requests to read/write 64KB data blocks with a steadily increasing address. There are from one to four clients (Workers in IOMeter’s terminology). Each client processes its own disk space range (there is a difference of 8GB between the ranges). The depth of the outgoing request queue is varied from 1 to 8.

I built diagrams for the case with the shortest request queue. It is the most common situation in practice.

So, multi-threaded reading comes first.

The Western Digital is generally slower than its opponents when working via USB 2.0 and FireWire 400. But when connected to a FireWire 800 port, this drive leaves the others behind, especially when there is only one hypothetical user. The advantage is smaller when there are more clients accessing the drive.

Now, let’s see what we have at multi-threaded writing.

It’s somewhat different than in the previous test: the Western Digital is the slowest with USB 2.0 and second with FireWire 400. When connected via FireWire 800, the WDG1T5000N leaves no chance to its opponents at any load – it knows no rivals when it comes to multi-threaded writing.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 ]

<%BANNER[banner_468x60_f]%>

Discussion

Comments currently: 4
Discussion started: 03/14/07 02:05:01 PM
Latest comment: 01/12/08 06:06:45 PM

View comments

You must log in to add comments.

Forgot password? Registration

remember me