Removable disk problems
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: <Jim.Lovell_at_email.protected>
Date: Tue, 22 Aug 2006 14:40:21 +1000
Hi everyone,
In getting ready for the experiment tomorrow I've discovered some problems with the removable disks that are quite serious. As Tid is the only station using removables this time this doesn't affect anyone else for tomorrow but I'd suggest that everyone runs some tests on their removable drives soon.
In the process of installing and testing the disks I ran Chris' diskspeed program on individual disks and found some had good data rates (~355 Mbps) while others were much worse (~160 Mbps in some cases). In a four disk RAID configuration I was getting rates anywhere between 200 and 900 Mbps and could see that some disks were taking longer to write than others by looking at the red lights on the front of the caddies. I also noticed that if I gave the disks a little push I could increase the data rates! This seemed like a connectivity problem to me. Anyway, last night, with the disks in their caddies and good looking rates, I ran diskspeed until the disks filled up. The resulting data rate (Mbps) vs time (s) is in test.gif. Note the horrrible performance! The configuration couldn't sustain 256 Mbps at some times!
So this morning I took apart the caddy system and connected the hard drives directly to the IDE cables, thereby removing the possible flakey connections. The result (for 7000 sec) is in nocaddy.gif and the difference is remarkable! (although I don't know what caused that slowdown around 1600-2600 sec).
So I'd strongly encourage everyone who uses the removable disks to run some data rate tests to see if they also have this problem!
For tomorrow's experiment I'll be using the disks in this hard-wired mode. The disk change will take longer but the data should be good.
Cheers,
Jim
with the
<div>
<img src="../../att-0603/test.gif" alt="test.gif
" />
Received on 2006-08-23 09:10:39
Date: Tue, 22 Aug 2006 14:40:21 +1000
Hi everyone,
In getting ready for the experiment tomorrow I've discovered some problems with the removable disks that are quite serious. As Tid is the only station using removables this time this doesn't affect anyone else for tomorrow but I'd suggest that everyone runs some tests on their removable drives soon.
In the process of installing and testing the disks I ran Chris' diskspeed program on individual disks and found some had good data rates (~355 Mbps) while others were much worse (~160 Mbps in some cases). In a four disk RAID configuration I was getting rates anywhere between 200 and 900 Mbps and could see that some disks were taking longer to write than others by looking at the red lights on the front of the caddies. I also noticed that if I gave the disks a little push I could increase the data rates! This seemed like a connectivity problem to me. Anyway, last night, with the disks in their caddies and good looking rates, I ran diskspeed until the disks filled up. The resulting data rate (Mbps) vs time (s) is in test.gif. Note the horrrible performance! The configuration couldn't sustain 256 Mbps at some times!
So this morning I took apart the caddy system and connected the hard drives directly to the IDE cables, thereby removing the possible flakey connections. The result (for 7000 sec) is in nocaddy.gif and the difference is remarkable! (although I don't know what caused that slowdown around 1600-2600 sec).
So I'd strongly encourage everyone who uses the removable disks to run some data rate tests to see if they also have this problem!
For tomorrow's experiment I'll be using the disks in this hard-wired mode. The disk change will take longer but the data should be good.
Cheers,
Jim
with the
<div>
<img src="../../att-0603/test.gif" alt="test.gif
" />
- application/octet-stream attachment: test.gif
- application/octet-stream attachment: nocaddy.gif
