I recently did some drive upgrades in my personal laptops and as a result I had a couple of SSD drives sitting around that I wanted to make use of.  A few weeks ago I purchased a new OCZ Agility 3 240GB SATA III SSD from Newegg when it was on sale and at the same time I also purchased two MassCool USB 3 external enclosures to make use of the two 120GB OCZ Vertex 2 SSD’s that I would have sitting around.  The cost for the MassCool enclosures was only $14.99 so I figured it would be a good way to reuse the SSD’s at the time.  When the package from Newegg arrived, I immediately installed the SSD into one of my personal machines and then later I went about testing the performance of the USB 3 external enclosures with the older SSD’s.  Initially the performance was impressive, and I posted a few tweets that attracted some attention and a few people requested that I blog my results.

To make this blog post something worth looking at, I ran a number of additional tests over the last few days using SQLIO and a varying configuration of drives with external enclosures.  Specifically I tested two different USB 3 2.5” SATA enclosures, a separate eSATA 2.5” SATA external enclosure, as well as two USB 3 HDD’s that I already owned to get some performance information from each of the setups.

The external USB 3.0 HDD’s being tested are as follows:

The external enclosures being tested in these tests are as follows:

The SSD’s used with the external enclosures listed above are:

At the time that I bought the 120GB drive I paid close to $2.42 per megabyte for it, and at the time this was a good deal.  When I purchased the 240GB SSD I paid right around $1.08 per megabyte, showing how much the prices have decreased nearly 3 years later.  I happen to own a number of OCZ drives at home and they are one of my favorite brands personally because of the overall reliability I’ve had with them the last three years as well as the performance to cost ratio I’ve experienced. Initially I was just testing the MassCool USB 3 enclosures that I purchased, and I was incredibly happy with the performance that I had from them.  However, after tweeting about the results, my friend Jose Chinchilla (Blog | Twitter) mentioned that I should also try out eSATA because it performed significantly better in his own tests.  Based on this tweet, I set out to my local MicroCenter computer store to purchase a eSATA enclosure and while I was buying it, one of the sales representatives, a guy named Chris, approached me and asked what I was planning to do with the enclosure because USB 3.0 should be faster for SSD’s. The numbers he quoted didn’t match my previous testing, so he made a recommendation for a specific USB 3.0 enclosure, and I figured, for $20 it was worth testing, so I bought an extra Hornettek Panther USB 3.0 device for comparison testing along with the Vantec NexStar CX eSATA enclosure I had already selected.

For the tests, I ran a short set of tests using SQLIO that I previously blogged about on my blog post about the Powershell parser for SQLIO output.  The results from the tests are below:

Looking at External Disk Performance using USB 3.0 and eSATA with SSD   image thumb

Looking at External Disk Performance using USB 3.0 and eSATA with SSD   image thumb

The two external USB 3.0 HDD drives had very similar performance characteristics, and they beat their USB 2.0 counterparts performance wise significantly.  I’ve been incredibly happy with my USB 3.0 HDDs overall for the last few years and I only purchase USB 3.0 HDD drives based on my performance tests a few years ago.  However, when compared with the USB 3.0 external enclosures with the SSD’s, the performance difference is quite significant.  One of the problems I’ve had lately is being able to fit all of the virtual machine hard drives, VHD’s for Hyper-V when using my dual boot Hyper-V host VHD, as well as my original VMware Workstation VMDK’s for classes, and even the VirtualBox virtual disk images (VDI) from my blog series on building a completely free playground.  Using the SSD’s with USB 3.0 definitely makes storing my virtual hard disk files on an external array much more feasible performance wise, and my only real limitation from testing is the size of the external device. 

However, the performance of the eSATA external enclosure is incredibly better, with the side trade off that I can only have 1 of them attached to my laptops at a time, and for my personal laptop, the fact that the eSATA enclosure requires not only the eSATA port for throughput, but also the additional USB 2.0 port for power really limits what I can do as far as having multiple disks connected to the laptop.  To be honest, this is something that I can live with given that I also have swapped out my CD/DVD ROM bay for a replacement New Mode US second HDD conversion. This means that I can have two 240GB OCZ SSD’s in my E5420, which is a similar configuration to the dual 256GB Dell SSD configuration in my Precision M6500 for work, while being able to move VM images between the two machines using my older 120GB SSD’s with fantastic performance.

If you are looking at really high performance external hard disk configurations for scalability, I would highly recommend looking at either eSATA with an extra SSD drive or if you need multiple devices and you have USB 3.0, consider going that route.  Either will outperform your existing options significantly.  For the time being, I am going to stick with one eSATA device, which is compatible with both of my laptops, and one USB 3.0 device, which only works with my M6500 at USB 3.0 speeds.  My E5420 only has USB 2.0 onboard, so it makes more sense for me to stick with eSATA for the main shared drive.