I used this Tripp lite 0.5M Internal Sata SFF-8484 To 4X7PIN Sas Cable with my Perc 6i and then ordered the rest from a chinese website for a fraction of the cost, but it took a few weeks to get here. I don't think mine are a crossover type.
8 x 1.5TB + Perc 6i + RAID-6 = Faster Storage
- Thread starter Stereodude
- Start date
I used this Tripp lite 0.5M Internal Sata SFF-8484 To 4X7PIN Sas Cable with my Perc 6i and then ordered the rest from a chinese website for a fraction of the cost, but it took a few weeks to get here. I don't think mine are a crossover type.
CougTek
Hairy Aussie
Will we be granted with another picture from Mr. Fingerman? :mrgrn:
Yeah it was the cables, those work much better. Hope I can return the others.
Did the default 64K, no read ahead, write back cache (I have a battery). Volume has whatever the default NTFS cluster size is. Running the bench now.
Did the default 64K, no read ahead, write back cache (I have a battery). Volume has whatever the default NTFS cluster size is. Running the bench now.
Also try adaptive read-ahead and see if you like that option. I have to check, but I think I'm using either 128K or 512K on mine.
I googled around and it seems it may get in the way of the read ahead in the OS if there is any. On an impulse I decided to just do everything default to start with. On the plus side there was no initializing time for a new array.
On the manager it looked really complete until I realized it didn't show the drive serial numbers. How the heck do you find the disk to pull out?
On the manager it looked really complete until I realized it didn't show the drive serial numbers. How the heck do you find the disk to pull out?
I googled around and it seems it may get in the way of the read ahead in the OS if there is any. On an impulse I decided to just do everything default to start with. On the plus side there was no initializing time for a new array.
On the manager it looked really complete until I realized it didn't show the drive serial numbers. How the heck do you find the disk to pull out, assuming one failed?
On the manager it looked really complete until I realized it didn't show the drive serial numbers. How the heck do you find the disk to pull out, assuming one failed?
Stereodude
Not really a
I'm pretty sure if you watch it you'll see it does a background initialization which will definitely affect your benchmark numbers.
Your right find it in the logs, background initialization started. Then found the separate window where you can track progress. I was looking and seeing the array status as optimal and thinking I was set for some dumb reason.
Apparently it goes through and checks the battery to learn the percent charge which is cool, see that in the logs (discharging/relearn). So what I see in logs is relearn in progress, battery discharging, battery capacity below threshold, and finally BBU disabled changing WB logical drives to WT. Guess I need to contact the seller eh?
Apparently it goes through and checks the battery to learn the percent charge which is cool, see that in the logs (discharging/relearn). So what I see in logs is relearn in progress, battery discharging, battery capacity below threshold, and finally BBU disabled changing WB logical drives to WT. Guess I need to contact the seller eh?
Stereodude
Not really a
Give it some time... That's normal. It should charge back up and re-enable WB.
I wish I looked for and read these threads earlier. Sure does require forced air cooling, sure enough it was pretty hot despite having a fan near it, painful to the touch even. I switched to having a fan right on it. Also the cabling note is in there. About 50% initialized now. Battery came up and reenabled the WB cache. Battery says 62% absolute charge and 99% relative charge whatever that means.
http://www.overclock.net/raid-controllers-software/359025-perc-5-i-raid-card-tips.html
http://www.hardwarecanucks.com/forum/storage/38530-dell-perc-6-i-firmware-upgrade-possible.html
http://www.overclock.net/raid-controllers-software/359025-perc-5-i-raid-card-tips.html
http://www.hardwarecanucks.com/forum/storage/38530-dell-perc-6-i-firmware-upgrade-possible.html
If you're interested in what I did. I have a cheap solution for cooling it which I mentioned in this thread. I'm using a Scythe 40mm fan to cool the chip. It's $7 and I attached it with nylon tie straps.
Other photos.
Other photos.
Stereodude
Not really a
I have a 80mm or 92mm fan blowing directly across the PERC card in each machine I have one in. You don't need a fan mounted directly to the HSF, but they definitely need direct airflow.
Stereodude
Not really a
You should increase the background init rate to 100%. The Init should only take about 6 hours.
Yeah that's what I did, I rigged up a bracket and aimed a 80mm fan at the heatsink.
I didn't see a rate for initialization but I did cancel and start over by using the fast initialization option. Seems to be moving faster than before.
I didn't see a rate for initialization but I did cancel and start over by using the fast initialization option. Seems to be moving faster than before.
Looking much better.
PERC6i in PCIEx4 slot
RAID6 with 64k stripe
Eight Seagate 5900RPM drives ST32000542AS
One big GPT partition with 4k clusters
Windows write cache looks like it is off with a message it can't be enabled, guess because the controller is doing it and the driver disables that?
Desktop pushing to server storage array with fastcopy. Same files as before from the other thread.
Wondering if I should switch to 64k clusters on the partition? I'm going to run atto and see what you guys think of those numbers.
PERC6i in PCIEx4 slot
RAID6 with 64k stripe
Eight Seagate 5900RPM drives ST32000542AS
One big GPT partition with 4k clusters
Windows write cache looks like it is off with a message it can't be enabled, guess because the controller is doing it and the driver disables that?
Desktop pushing to server storage array with fastcopy. Same files as before from the other thread.
Code:
TotalRead = 35006.1 MB
TotalWrite = 35006.1 MB
TotalFiles = 898 (37)
TotalTime= 446.13 sec
TransRate= [b]78.47 MB/s[/b]
FileRate = 2.01 files/sWondering if I should switch to 64k clusters on the partition? I'm going to run atto and see what you guys think of those numbers.
Stereodude
Not really a
Atto is basically worthless because the data size it uses is too small relative to the cache size of the drives and the cache on the PERC card.
I got my names mixed up, it is IOmeter but using this "Super ATTO Clone pattern.icf" configuration file. Still I'd agree with that, it ran for over four hours and did nothing. I have a several terabyte test file on the drive that it generated.
From the ICH9R mirror to the array:
In general I think the previous posted speed over gigabit is the best I can hope for which suits me fine.
From the ICH9R mirror to the array:
Code:
TotalRead = 35006.1 MB
TotalWrite = 35006.1 MB
TotalFiles = 898 (37)
TotalTime= 375.28 sec
TransRate= 93.28 MB/s
FileRate = 2.39 files/sIn general I think the previous posted speed over gigabit is the best I can hope for which suits me fine.
Stereodude
Not really a
IOmeter works best on a unpartitioned drive. If it has a partition it will fill the drive with a file that is the size you specify and use that file as the location for the data it reads and writes to the drive.
Yeah but my thoughts were I want to test different partition block sizes. Meh, anyway the thing seems more than fast enough for me.
Restoring backup 2
Restoring backup 3
Started out with files that were added last at about 65MB/s more or less. After it got off the folder that was added last it rose to 75MB/s and higher for the files that were added to the drive first. I conclude that the network/server can take files faster than the source drive can send them now. Which is in teresting in itself as both the array and the backup drives are the exact same Seagate 5900RPM drives.
Also for me I don't really see that fastcopy adds much over explorer copy paste speed wise.
Restoring backup 2
Code:
TotalRead = [b]1897714.8 MB[/b]
TotalWrite = 1897714.8 MB
TotalFiles = 21061 (1233)
TotalTime= 25533.36 sec
TransRate= [b]74.32 MB/s[/b]
FileRate = 0.82 files/sRestoring backup 3
Started out with files that were added last at about 65MB/s more or less. After it got off the folder that was added last it rose to 75MB/s and higher for the files that were added to the drive first. I conclude that the network/server can take files faster than the source drive can send them now. Which is in teresting in itself as both the array and the backup drives are the exact same Seagate 5900RPM drives.
Code:
TotalRead = [b]1887379.4 MB[/b]
TotalWrite = 1887379.4 MB
TotalFiles = 13791 (969)
TotalTime= 24456.36 sec
TransRate= [b]77.17 MB/s[/b]
FileRate = 0.56 files/sAlso for me I don't really see that fastcopy adds much over explorer copy paste speed wise.