Search
Sign in with
Sign up | Sign in
Solved

RAID 1 with SATA and IDE drive

Last response: in Storage
Share

Hello,

I have a VIA PCI Raid card that supports both SATA and IDE. I have created a RAID 1 with a Veloci Raptor 160gig sata drive, and a Seagate Baracuda 160gig IDE drive.

I realize the SATA drive is much faster, but they way I understand RAID 1 is that data is read from or written to the Source disk and Mirrored disk at the same time, but as soon as the source disk receives the write the OS receives the notifications. This could be wrong.

Obviously in most RAID types, your slowest drive is going to be the speed at which your system operates. I was thinking that in RAID 1 as long as your speed is limited by your "source" disk. Please correct me if I'm wrong..

More about : raid sata ide drive

Register or log in to remove
Storage Expert

If I had to bet, I would bet that you are wrong, but it would only show up under heavy write loads. Small loads get buffered and you only see the effect of the buffer.

Grab a good benchmarking tool and get serial write rates for the array, then compare them to serial write rates for each drive (from a review or the mfg site). Then let us know what you see, and we'll all learn something.

I used a program called "HD Speed" it only test read speed. It only sees one of the drives in the array as a "physical drive" then it listed the "C: drive" which would be the ARRAY. Both had read speeds of around 101.0 MBps
Related ressources
Ask the community

Best solution

Storage Expert

For a RAID-1 set, decent controllers usually issue reads to the drive whose heads are closest to the requested data. This means that read performance of the RAID set should be better than the performance of the IDE drive alone - in other words, read performance doesn't degrade to that of the lowest common denominator.

Write performance is always worse than an individual drive with RAID-1 because even though the controller can issue both writes in parallel, it still has to wait for the slowest drive to accept the data before it can report that the write has completed. If your RAID controller has battery backed-up RAM then it can report completion of the writes immediately while caching the data until the drives are actually able to accept it. That works very well for light-duty writing, but if you're writing a large volume of data then the cache will probably fill up and reduce the write performance to that of the slowest drive.

Snipergod87 said:
I believe that it will only go as fast as your slowest drive. Benchmarking will help us findout.





You are correct. A RAID-1 array will only operate as fast as the slowest disk in the array.

mavroxur said:
You are correct. A RAID-1 array will only operate as fast as the slowest disk in the array.


+1 RAID1 is mirror and no part of the mirror can be faster than another.

Well I deleted the array and I can now see the drive independently. I repeated the read test and the Velociraptor reads at 101MBps and the Seagate Barracuda reads at 66 MBps. So it looks like with my RAID 1 setup the read speed was indeed that of the faster drive.. however I'm betting the write speed would have gone the way of the lowest common denominator.. If anyone knows of a free program that tests write speed I can double check that..
Storage Expert

mavroxur said:
You are correct. A RAID-1 array will only operate as fast as the slowest disk in the array.


misry said:
+1 RAID1 is mirror and no part of the mirror can be faster than another.


Just to head off any misconceptions, these statements are true for writing but not necessarily for reading. When reading, if there are multiple outstanding I/O requests then both drives can be read in parallel - therefore under ideal conditions you can get up to twice as much throughput out of the array (or the sum of both drives' individual throughput in the case of drives whose performance differs).

This isn't generally as big an advantage for desktop systems most desktop applications will tend to issue I/Os sequentially rather than in parallel, but it can have a big impact in server applications.
Storage Expert

stephenspann27 said:
If I got a PCI (not express) sata 3.0 card, would I see much of an increase in speed? I'm guessing this drive is only running at Sata 1 speed.
No. Even a 1.5Gbit/sec SATA connection is already faster than any of your drives, providing a fatter pipe from the drive to the system isn't going to improve the speed of the drive itself. The only way to make the drive itself faster is to increase its spin rate or cram more data around each track - not something you can do except by using a different manufacturing process to build a new, improved drive.

The computer I have it installed in is a Dell Dimension 4600 (yr 2004). I got about 96.6MBps for a read speed when using my VIA PCI Raid card, and I get around 102MBps when I have it plugged directly into the motherboard. That's a typical speed for a veloci-raptor drive? If I were to plug it into a brand new motherboard I would see the same speed?
Register or log in to remove
 
Ask the community
Read discussions in other Storage categories
!