Hard drive performance

[Dave2002]

I'm just wondering what is considered a reasonable or good rate for data transfer to an external drive these days. 1 Gbyte/minute perhaps?

I have just bought a multi standard drive in order to aim at a higher throughput. To attempt to make that work one of the interfaces is FireWire 800, since the particular iMac I'm trying to use doesn't have any USB3 capability. I think the Firewire drive has achieved around 1 Gbyte/minute earlier this evening, as measured very crudely.

My drive is something like this one - http://www.ebuyer.com/271704-freecom...2259&pkw=&pmt= - though that seems to be a good price - I paid slightly more.

Looking at various reviews, one caught my eye - for an empty enclosure with Firewire 800 and USB 3 - and a review suggesting that it worked well with a Crucial SSD. The total cost might have been around £150, but the claim was that it was very fast, and as fast in use as booting directly off the main computer drive.

I suspect that most drives of whatever variety do not get close fo the theoretical limit of their performance. I'm assuming that my ageing iMac is fundamentally limited for data transfer to and from storage, with only USB 2 and Firewire 800 worth considering, though the USB could be used with USB 3 devices, but would not achieve USB 3 levels of throughput. My hunch is that Firewire 800 will be the fastest, but that might turn out to be incorrect.

[Anastasius]

Try this article http://uk.pcmag.com/hard-drive-revie...rives-for-macs

or this article http://www.macworld.co.uk/feature/ma...e-mac-3600870/

I think that if you Google enough you will find that someone has already done the heavy-lifi=ting for you.

[Dave2002]

Try this article http://uk.pcmag.com/hard-drive-revie...rives-for-macs

or this article http://www.macworld.co.uk/feature/ma...e-mac-3600870/

I think that if you Google enough you will find that someone has already done the heavy-lifi=ting for you.

I don't have accurate results yet for my Freecom drive - for various reasons, but so far the best rate I've had off the new drive is 5 Gbytes/minute via the Firewire interface. That's around 660 Mbps, which is a higher rate than USB 2. An earlier comparison of a USB 3 drive compared with the Firewire showed them on a 10 Gbyte test to be quite similar - in about 3-4 minutes, which works out at somewhere in the range 333-444 Mbps.
I did have a hunch that on the same test the Freecom drive with Firewire won - though there wasn't too much in it. I think a lot depends on the particular job to do. Transfer rates seemed significantly higher with a lot of modest sized files, rather than a few really big ones - perhaps there is an OS reason for that.

The Freecom drive is very sturdily built - quite solid and heavy, BTW.

It's early days yet, and mainly I'm using the drives I've got to try to put my 2009 iMacs in order. Obviously newer ones (say from 2012 onwards) would be likely to have USB 3 and or Thunderbolt, which I would expect to be significantly faster, but given that I've been trying to put new life back into these older machines, the few £00s I've spent so far on memory upgrades and a few hard drives compare quite favourably. The cheapest new machine I'd want to buy would probably cost about £1500, and I could be tempted to spend more!

I am now gradually moving data on to the external drives, and trying to get reasonable working space on the main machines. Hopefully things should get easer as the space margins increase.

I also tried to do a two way data transfer comparison using the Firewire drive, as it should be capable of doing full duplex operation, though so far that hasn't been shown to be ultra fast. It would take more care and effort to set up a good test for the FIrewire drive and interface. It could be that even though the drive can do full duplex, there is some interaction between the up and down transfers, so it's not necessarily going to be faster. In theory, if full duplex does work well enough, then the uploads and downloads can be overlapped, to almost double the throughput and nearly halve the overall work time.

I have wondered about Thunderbolt drives, but they are/and have been expensive, and some of my machines can't do Thunderbolt anyway.
Getting in these drives which still use the older Firewire interface is an attempt to avoid having to discard machines which otherwise work perfectly well.

Lastly, yet again amazing service from Crucial. Another upgrade kit ordered yesterday turned up through the letter box today.

[Anastasius]

I'm not sure I'm interpreting your statement regarding transfer rate of large numbers of small files vs a few large files correctly....it can be read either way!

General understanding and observation in the field is that it will take longer to transfer 1000 files of 1Mb each compared to one file of 1GB because you have the additional overhead of the file management involved in opening and closing files. But the whole issue of timing the transfer of files is dependent on so many, many variables. There is a lot of information out there. https://discussions.apple.com/thread...art=0&tstart=0 as just a starter.

[Dave2002]

I'm not sure I'm interpreting your statement regarding transfer rate of large numbers of small files vs a few large files correctly....it can be read either way!

General understanding and observation in the field is that it will take longer to transfer 1000 files of 1Mb each compared to one file of 1GB because you have the additional overhead of the file management involved in opening and closing files. But the whole issue of timing the transfer of files is dependent on so many, many variables. There is a lot of information out there. https://discussions.apple.com/thread...art=0&tstart=0 as just a starter.

I thought this was pretty clear (apart from a typo, now corrected) and unambiguous -

Transfer rates seemed significantly higher with a lot of modest sized files, rather than a few really big ones - perhaps there is an OS reason for that.

It's not by any means obvious that transferring a few large files would be faster than transferring a lot of smaller ones. True, there would be some overhead in looking up the details for each small file, but that could be very fast. You are right that there may be many factors. Other influences might be the way that large files are organised on drives or in memory, and the available memory used for buffering might also have an effect. There may be "general understanding", but it could very well be incorrect. Also, different people using different systems may genuinely have experiences which do not match.

The link you mentioned had some views, but they weren't particularly helpful to me. There was confirmation that there can be more factors to influence the transfer rates. I have no intention of daisy chaining FW drives, or using FW400. The rates mentioned generally seemed very slow, as I've already experienced much higher rates, and I think I could repeat that if pushed.

I generally work on the assumption that "real world" activity will not perform anywhere close to some theoretical design specs, but actually I think in my case the rates are not orders of magnitude out from what might be expected, and they are acceptable.

[Anastasius]

I'd be interested to know if your results were replicated between case 1 (large number of small files) and case 2 (one large file). Are you absolutely sure that when you ran these timing checks that the status of your Mac was the same? No background tasks running for one case but not the other ? Of course, you are entitled to dismiss 'general understanding' but then that would fly in the results from my thesis when I did my MSc in Computer Science on disk performance.

Your timing results are unique to your system as it stands and no inference can be made that that interface is better than that interface or that 'this' is better than 'that' because you are not starting from Ground Zero. As I have mentioned in many other threads that touch on this issue, until you have taken some baseline performance metrics of your drive using a program like Scannerz then the phrase implying expressing water while under inclement weather conditions comes to mind. You can have bad sectors that although readable take longer to read then they should. This will directly affect your timings. If some of those bad sectors are part of the large file then that could explain the results that you got. True - you have a brand new drive but unless you've done a baseline test on all the sectors on that drive, you can only hope that they are perfect.