eSATA RAID 1 as slow as FW800 (non RAID)??

[SpaceJello]

Got my hands on a friend's external drive and was testing some transfer rate, about 8 G or so of files. It should be sequential read right? So eSATA would be faster... I think.

But why was the transfer speend the same approximately as the transfer to a FW800 external drive? Is it because it's RAID 1? and its mirroring?

Thanks!

 

[CanadaRAM]

It's likely because you are hitting the limit of the drive mechanism's transfer capability sooner than you are seeing a difference between eSATA and FW800.

Is the RAID 1 managed by a hardware controller in the drive enclosure, or is it software RAID? RAID 1 does have some overhead.

Also, you can't compare transfer rates on 2 different hard drives. If one of them is 90% empty, and the other is 90% full, the full one might perform as much as 50% slower. Not to mention if there are differences between the mechanisms.

 

[SpaceJello]

It's likely because you are hitting the limit of the drive mechanism's transfer capability sooner than you are seeing a difference between eSATA and FW800.

Is the RAID 1 managed by a hardware controller in the drive enclosure, or is it software RAID? RAID 1 does have some overhead.

Also, you can't compare transfer rates on 2 different hard drives. If one of them is 90% empty, and the other is 90% full, the full one might perform as much as 50% slower. Not to mention if there are differences between the mechanisms.

Actually the eSATA is the one that almost 99% empty, whereas the FW800 was like 90% full. I am not sure if its controlled by the drive enclosure?

This is the ony me friend had for the eSATA enclosure:
http://eshop.macsales.com/item/Other World Computing/MESATATBEK/

I was testing it with a WD mybook pro .

*Edit*
its using software RAID 1 according to the site. Does it matter, whether its software or hardware?

 

[sndcj1]

Well, my understanding is that software RAID uses the CPU to handle the RAID logic, so that would be slower than a standard drive by just a little. As for firewire 800 vs eSATA with 1 drive, or RAID1, the drive is going to be the bottleneck, with the possible exception of a cached read/write. According to Tom's Hardware most 7200 drives read at about 60 MB/s average read, or 480 Mb/s, well under either protocols data limits.

 

[hayduke]

According to Tom's Hardware most 7200 drives read at about 60 MB/s average read, or 480 Mb/s, well under either protocols data limits.

Agreed. I think a lot of people don't realize this about write speeds. Many set-ups are hard drive limited.

 

[SpaceJello]

Cool thanks, so i guess until we get affordable 10000 rpm drives, we aren't going to see the real benefits of higher speed connectors

 

[velocityg4]

Cool thanks, so i guess until we get affordable 10000 rpm drives, we aren't going to see the real benefits of higher speed connectors

Not really, since drives increase in speed due to data density as well. Most recent drives hold 160GB to 200GB per platter. Just a few years ago hard drives held 20GB to 30GB per platter and much lower. So even though the drives are still 7200RPM the increase in data density means that more data passes by the read/write head per revolution. Thus the an increase in data density improves read/write performance without increased RPM's.

10000RPM or even 15000RPM would be welcomed.

If you look at the price decrease and performance increase in flash hard drives over the last year spindle hard drives could be outmoded eventually. The new 2.5" 64GB solid state disc from SanDisk smokes current top of the line 2.5" spindle drives in read transfer rates and read latency. With write transfers still needing some catch up work though they are up about 70% over a year ago.

http://www.tomshardware.com/2007/08/13/flash_based_hard_drives_cometh/