Performance
With everything connected, drivers installed and the e.SATA kit powered up, I ran a bunch of well-known benchmarks to gauge the performance of the Highpoint e.SATA kit:
SiSoft Sandra 04
ATTO Disk Benchmark
HD Tach 2.70
The crucial specs for the review system where:
Intel 875PBZ motherboard
Pentium 4 3.2 GHz Extreme Edition
GeIL 512MB PC3200/DDR-400 Golden Dragon at SPD setting
Albatron Nvidia GeForce FX5900PV
Two 36GB WD Raptor 10Krpm SATA drives, RAID-0
For the Highpoint e.SATA kit, two 80GB 7,200 rpm PATA EIDE ATA-100 drives were used, also from Western Digital.
The operating system was Windows XP Professional SP1 with Windows Update patches and I formatted the hard drives in the e.SATA kit with NTFS using the default cluster size.
When you test hard drive performance in RAID-0 configuration, the "stripe size" or the size of the block of data that files are broken into when writing to the hard drives in the array has a large effect on the performance you'll see. However, picking the right stripe size isn't easy: the best size for your application (as in average file sizes, file system, disk platter numbers and density, hard drive and OS caching, plus the controller) is usually not the same as for another application, where just one of the variables differ. I'm afraid that it means only one thing... you need to try several stripe sizes to find the "sweet spot" for your particular application. The stripe width was two of course, corresponding to the number of drives in the array.
Therefore, I ran the RAID-0 tests with 16KB (smallest), 256KB (medium) and 2MB (largest) stripe sizes. The Highpoint tools for Windows made it quick and easy to create and delete arrays, and for comparison, I also tried the Windows XP software-only RAID-0 (Windows doesn't tell you the stripe sized used for that, unfortunately).
Sandra Benchmark
Look at that -- the external e.SATA kit performs as well as many internal RAID-0 set-ups, with SATA-150 drives. This is with the 16KB stripes; the buffered and sequential read figures are excellent.
With the 256KB stripe size, the performance is still very good, but while write performace wasn't affected as much, read performance took a severe hit; access time increased a little too, but that figure may be within the benchmark's margin of error.
The 2MB stripe size shows similar overall performance to the 256KB one, with the exception of the access time being slashed in half.
So, for big Sandra 04 Drive Index figures with the Highpoint e.SATA, you want small stripe sizes :-).
Windows XP software RAID-0 looks good in terms of overall performance. Random reads aren't too hot though.
ATTO Disk Benchmark
16KB 256KB 2MB
ATTO didn't like the 256KB stripe size much, and it was only with transfers sizes above 512KB that the 2MB stripe showed figures as good as those obtained with the 16KB stripe.
Again, the Windows XP software RAID-0 worked about as well as the Highpoint one, with the smallest cluster size. The exception was read performance, which didn't look as good with smaller transfer sizes.
HD Tach 2.7
e.SATA RAID-0 performance
The averages are about the same, but look at how the sequential speed fluctuates at the larger stripe sizes. Burst read speed (ie from the controller interface) is the highest with the smallest 16KB stripe size.
RAID-1 performance
As some people may want to use the e.SATA kit as an external back-up device, I tested with the drives in RAID-1 configuration too. This means the data is mirrored on both drives, and if one fails, you can carry on operating with the functional drive and replace the failed one and rebuild the array. RAID-1 provides a good measure of fault tolerance (especially if you were to mirror the controllers as well), but of course, as data is written to two disks, you cannot expect the blistering performance of RAID-0.
RAID-1 also gives "half" the available space of RAID-0, so the array shows up as 80GB to the operating system instead of 160GB.
Benchmarks all obtained with the drives configured through the Highpoint Windows management utility.
Conclusions
Well, there's no denying that the Highpoint e.SATA kit is a top performer. Even though it uses a PCI controller only which is theoretically limited to 132MBps bandwidth that is shared with other devices as opposed to Intel's ICH5-R that bypasses the PCI bus for a much faster 266MBps connection, the PATA drives in the e.SATA kit put in sterling numbers in the benchmarks. The Highpoint e.SATA kit is also relatively easy to install, and given how cheap IDE drives are nowadays, lets you create large external storage for not much money.
It's ideal if you don't have room inside the case for further drives and you also get two internal SATA connectors as a bonus. Pricewatch.com reckons Newegg has the e.SATA kit for $212 which isn't too bad overall.
However, the e.SATA kit isn't the answer if you want hot-pluggable portable external storage, like the USB 2.0 and FireWire IEEE-1394 solutions on the market currently. Other than that, it's a neat high-performance solution if you need to extend your storage options using cheap IDE drives.
Pros:
Excellent performance
Ventilated drive enclosure
Reasonable price
External PSU
Hot Swappable
Cons:
A few rough edges in the design of the drive caddies
Software and manuals need polish
Not easily portable