Benchmarking VM Fusion storage

I use a Macbook Pro with VMware Fusion, which lets me run a bunch of virtual Windows machines for testing. I upgraded my internal drive to a quick 320gb one a few weeks ago, intending to run all my VMs off that internal drive, making it easier to do testing on airplanes or in coffee shops. Problem is, one of the things I test is Quest LiteSpeed, database backup software that does a lot of disk I/O, and during heavy backup/restore testing, I couldn’t do much multitasking because the internal laptop drive was being hit so hard.

I figure it’s a matter of time before I put a second hard drive in the Macbook Pro, but for now, I’ll stick with external drives.

I’ve got a bunch of USB drives, so I did some comparisons to see which ones got the fastest storage performance inside VMware. I wanted to find out how much performance I’d really gain by switching to an external drive, and did it make a difference whether I used FireWire or USB.

Keep in mind that this is not the native performance of each drive – this is the performance as seen inside a VMware Windows XP guest. I didn’t care what the native performance is, because that doesn’t do me any good. I’m only interested in performance as seen by the VMware guests because that’s the only thing I’d use the external drives for.

Here’s the benchmarking results from HD Tune:

Internal Western Digital 320gb 2.5″ WD3200BEVT:

WD3200BEVT

Big, bulky external Maxtor OneTouch III 3.5″ 500gb connected via FireWire 400:

External 500 via FireWire 400

Whoa. That’s only averaging 35.5 MB/sec, whereas the internal gave me 48.5 MB/sec. I’m surprised there.

External Maxtor OneTouch III 3.5″ 500gb connected via USB:

External 500 via USB

Not much difference in speed between FireWire and USB as far as speed goes, but the USB connection used almost double the CPU power.

External USB enclosure with a Fujitsu MHY2120BH 2.5″ 120gb (the Macbook Pro’s original internal drive):

External 120 via USB

I’ve also got an external SATA RAID enclosure that does mirroring & striping, and I’ll test that later just out of curiosity, but there’s no way I’d put my virtual machines on there because it weighs more than my laptop and it makes a loud racket.

These results are not scientific – I just did one pass of testing on each drive. Your mileage may vary. Offer not valid in all fifty states. No purchase required to win. See participating locations for details.

My verdict: I’ll put the virtual machines on the 2.5″ external enclosure.  It’s tiny, doesn’t require external power, and it won’t be as fast as the internal drive – but at least it’ll let me multitask.

Previous Post
Performance troubleshooting flowchart
Next Post
Getting Into a Database Administrator Position

2 Comments. Leave new

  • Hey thought I would try HDTune on my macbook pro internal drive because I paid an extra 200 bucks or something for the internal 200GB SATA @7200 rpm. I have the screenshot but it shows I’m averaging 72.9 MB/sec. I did run it inside Windows XP running in Fusion on the macbook. One question I have is it shows CPU usage at 56.6% which is way higher than yours. Is that an indication I have something configured wrong in Fusion?

  • Ok tried it again with all unnecessary applications closed on the Host and Guest OS’s. This time I got a 90.9 MB/sec average transfer rate and CPU Usage of 13/4% which seems a lot better. Might be an obvious item but I’m not sure how having more applications opened on the host or guest would increase the CPU Usage %.

Menu
{"cart_token":"","hash":"","cart_data":""}