The DAM Forum
Welcome, Guest. Please login or register.
May 21, 2013, 04:03:14 AM

Login with username, password and session length
Search:     Advanced search
Jan 9, 2012
John Beardsworth's new Lightroom site
Lightroom Solutions
27960 Posts in 5113 Topics by 2914 Members
Latest Member: imthedamstar
* Home Help Search Login Register
+  The DAM Forum
|-+  DAM Stuff
| |-+  Hardware Discussions
| | |-+  hardware vs software RAID (Windows XP)
« previous next »
Pages: [1] Print
Author Topic: hardware vs software RAID (Windows XP)  (Read 1550 times)
stuhenderson
Newbie
*
Posts: 3


View Profile
« on: September 29, 2006, 09:41:05 AM »

Any suggestions or opinions on doing RAiD in hardware or software?  Windows XP seems to support it in the software, but I have no experience with it.  I'd appreciate any opinions, experiences, advice.  Thanks very much.  Stu
Logged
Marc Rochkind
Hero Member
*****
Posts: 1136


View Profile WWW
« Reply #1 on: September 29, 2006, 09:51:52 PM »

One caution I would offer is that RAID is not backup, chiefly because it doesn't offer enough separation. Theft, fire, surge, flood, hardware failures, and like would still affect the whole array. There's no performance improvement either--more likely, a degradation. RAID, depending on how it's configured, can provide very rapid recovery from crashes. But, for imaging work, this isn't nearly as important as it is for transaction-based systems and other similar systems that need to provide high degrees of availability.

--Marc
Logged

ralphpaonessa
Full Member
***
Posts: 108


View Profile WWW
« Reply #2 on: September 29, 2006, 10:20:52 PM »

There's no performance improvement either--more likely, a degradation.

What about RAID 0, where data is "striped" across 2 or more disks simultaneously? Does this provide any performance advantage when reading/writing to disk in a single-user system?

What is the limiting factor when reading/writing to disk. E.g., I have five Western Digital SATA disks on my Dell Precision 360 Workstation (Windows XP). I recently tried copying several gigabytes of data from disk to disk and timing the process. Basically, I got transfer rates around 30-40 MB/sec. The disk specs say the maximum transfer rates are:

Buffer to host (SATA): 300 MB/sec
Disk to buffer: 748 Mbits/sec = 93.5 MB/sec
Buffer size: 16 MB

I'd hoped that transfering large blocks of files from disk to disk should occur close to the 93.5 MB/sec, but I only got about a third of that. What other bottleneck(s) are limiting the transfer rate?

Are there any practical steps to take to speed this up? E.g. upgrade to latest Pentium Dual Core systems?

All my drives are Western digital. The newest are "SATA 300," the older ones are "SATA 150," and one (the boot drive) is a 10,000 RPM Raptor drive. Two are internal with motherboard SATA connections. Three are connected to a new external box via a SATA Multilane cable, which connects to an Addonics SATA card I just bought. But there's no dramatic difference in my simple testing on the speed of moving data from drive to drive.
Logged

Ralph Paonessa
RPPhoto.com
Marc Rochkind
Hero Member
*****
Posts: 1136


View Profile WWW
« Reply #3 on: September 30, 2006, 06:43:52 AM »

I can't answer most of your questions, but I'd be surprised that disk performance would be a factor in imaging work. Perhaps you are right that RAID 0 provides a performance improvement. There are numerous factors inside the OS and inside apps that affect overall disk performance besides the properties of the drive itself.

--Marc
Logged

stuhenderson
Newbie
*
Posts: 3


View Profile
« Reply #4 on: September 30, 2006, 08:39:46 AM »


There's no performance improvement either--more likely, a degradation.
Quote

What about RAID 0, where data is "striped" across 2 or more disks simultaneously? Does this provide any performance advantage when reading/writing to disk in a single-user system?

What is the limiting factor when reading/writing to disk. E.g., I have five Western Digital SATA disks on my Dell Precision 360 ...


The performance advantage from RAID 0 comes from having two disks writing at once, so (with some loss to side issues), the writing of a file should take about half the time.  Replacing the CPU chip will have no effect, since this is arithmetic and logic are performed, not the writing to disk.

Your speed loss may come from having RAID 0 on different model drives.  Disk drives spin pretty fast, but if you want to write at a given place on the disk, you have to wait for the drive to spin to that place.  With different type drives, the spinning will be at differernt speeds, which might cause one disk to have to wait for the other to finish.  Is it possible to re-configure your RAID 0 to write to two identical model disks and then see the timing?  Hope this helps.  Stu



Logged
ralphpaonessa
Full Member
***
Posts: 108


View Profile WWW
« Reply #5 on: September 30, 2006, 11:30:51 AM »


What is the limiting factor when reading/writing to disk.

After asking this question, I found the following extensive guide on Hard Disk Performance. Everything you ever wanted to know and more (although not up to date on the latest disk technologies). After browsing this, I can see that many factors combine to determine disk performance in practice (numerous disk specs, computer bus, particular types of demands the application makes on the disk). The author's practical bottom line seems to be: "Buy a relatively recent, relatively high powered system with recent, relatively high powered disks, and don't obsess too much over the disk specs."

Just the excuse I needed to upgrade to Dual Core Pentium!
Logged

Ralph Paonessa
RPPhoto.com
Pages: [1] Print 
« previous next »
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!