 |
|
|---|
The Promise SuperTrak SX6000
|
Few argue that Promise has enjoyed greater success in the ATA RAID market than any other manufacturer. In particular, their cost-sensitive FastTrak cards are by far the most popular and well-known of all. Promise's SuperTrak series, on the other hand, targets the much smaller high-end ATA RAID market. Target markets shared with 3ware's Escalade series and Adaptec's ATA RAID 2400A, the SuperTrak series lives on with its newest member: the SX6000.
The SuperTrak SX6000 is the successor to the SuperTrak100 reviewed earlier this year. The ST100 was a huge disappointment from a performance standpoint, performing no better in any RAID level relative to the performance of a single drive. With the SX6000, Promise obviously hopes to improve immensely on the ST100's performance. The SX6000 brings many features that one would expect from a high-end ATA RAID card, including on-board cache and RAID 5 support. Its full specs are as follows:
Some notes about the card's specs... Firstly, its on-board processor is "new and improved" relative to the ST100's. Secondly, the SX6000 is, to our knowledge, the first RAID card (or ATA controller in general, for that matter) to support 48-bit LBA. This allows the full capacity of drives bigger than 137GB to be utilized (28-bit LBA can only address 137GB worth of sectors). Thirdly, the card's claimed support of RAID 3 isn't accurate: because the only available stripe size range is 4K-64K, a "RAID 3" array on the SuperTrak SX6000 is, in reality, a RAID 4 array. See our RAID guide for more information on the differences between RAID 3 and RAID 4. Lastly, the adapter requires a minimum of 16MB of memory, despite the fact that it doesn't come with any. This may seem odd at first, but there is indeed some logic behind this decision. A quote from Promise's FAQ:
How come you don't include memory on the SuperTrak SX6000?
"The user is free to choose exactly how much memory he or she wants to install, and which type to use. Additionally, it is much more cost-effective for the user to purchase memory from a retailer rather than paying the additional amount that Promise would need to charge for the card if it included memory." This makes sense, as the price mark-up on RAM sold by companies such as RAID card manufacturers tends to be extravagant. (Memory upgrades for some of Adaptec's SCSI RAID cards, for example, are priced upwards of 10 times higher than standard SDRAM.) It's also worth pointing out that the SX6000 is apparently much more tolerant of different SDRAM types/manufacturers than other cards such as the ATA RAID 2400A. The retail SX6000 kit includes the following items:
The SX6000's BIOS is your run-of-the-mill RAID card BIOS, permitting standard functions such as array creation, deletion, and rebuilding. The SuperTrak SX6000 comes with a utility known as PAM, which stands for Promise Array Management. PAM allows arrays to be created, deleted, and rebuilt from within Windows. Arrays can also be monitored, both locally and remotely. If PAM detects a problem with an array, it can automatically send an email to the appropriate email address. The user may choose exactly which events should and shouldn't trigger an email, such a bad drive, bad sectors, an automatic rebuild, etc.
The following benchmarks were run with the SX6000's read/write cache enabled and set to 'write-back', and a stripe size of 64K. Driver release 1.10 (build 31) was used along with BIOS version 1.10 (build 10). The card was tested with a 32MB SDRAM module.
Base Scores...
Nothing out of the ordinary here - scores are what one would expect. Note, however, that the SX6000 doesn't suffer from the unfortunate 20MB/sec STR limitation that the ST100 did. Good news, indeed.
The SX6000 comes in a bit high on the access time side of things, but the difference may not be significant. The SX6000 seems to be limited to about 70MB/sec worth of STR, which, while lower than that of the competition, is quite an improvement relative to the ST100.
Note that the SX6000 doesn't load-balance sequential reads in RAID 1 like the Escalade does. Hopefully this isn't an indication that the card doesn't load balance random reads either...
The SX6000 weighs in with significantly lower access times in this test, a good sign. STR just about matches the competition.
Since the ST100 is the only other ATA RAID adapter besides the SX6000 to support RAID 3, there's not much to compare the SX6000 to. However, the SX6000's STR in RAID 3 is much higher than that of the ST100 - hardly surprising given the STR results above.
The SX6000 again arrives on the high end of the access time curve, but we'll wait to see what IOMeter has to say before drawing any conclusions.
Base Scores...
What stands out from the above scores is the SX6000's poor write performance - both random and sequential. Performance in our core tests (File Server, Workstation, and Database), however, is as expected. RAID 0...
As the above graphs demonstrate, the SX6000's performance is remarkably improved relative to the ST100. Scores in our core tests scale nicely as drives and outstanding IO's increase. Core test scores remain lower than those of the Escalade 7410 and ATA RAID 2400A, however, particularly in the Workstation pattern. Write scores are disappointing - for whatever reason, the SX6000 seems to be limited to about 4.5MB/sec of STR per drive. Random write performance isn't anything to boast of either, unfortunately. RAID 1...
Like the Escalade 7410 and ATA RAID 2400A, the SuperTrak SX6000 boosts RAID 1 performance by load balancing random reads between both drives - a positive thing, indeed. Scores aren't quite up to the level of the competition, however, and write performance continues to be poor. RAID 01...
Scores again greatly improve upon the ST100, but remain behind the competition. RAID 3...
Unsurprisingly, the SX6000 does a lot better than the ST100 in RAID 3. RAID 5...
In RAID 5, the SX6000 does particularly well against the Escalade 7410, topping 3ware's offering in most tests. It performs significantly worse than the ATA RAID 2400A, however, which continues to be the ATA RAID card to beat in RAID 5.
When trying to summarize the SX6000's performance, the phrase "it's all relative" works quite nicely. On one hand, we can't help but be impressed with the SX6000's improved performance relative to the ST100 - we couldn't help but be somewhat worried prior to testing at the possibility of the SX6000 performing anywhere nearly as poor as the ST100. Thankfully, this didn't turn out to be the case. Relative to competing controllers, however, the SX6000's scores don't compare quite so favorably. Scores in RAID 0, 1, and 01 are significantly lower than the Escalade and ATA RAID 2400A, particularly in the write tests where, for whatever reason, the SX6000 does very poorly. RAID 5 is somewhat of a different story... Who'd of thought that the successor to the sickly ST100 would perform better overall in RAID 5 than the Escalade 7410? Not very many, we suspect. It can't match the performance of the ATA RAID 2400A, however, although it does offer the advantage of supporting up to 6 drives as opposed to the 2400A's limit of 4. In the end, it's not out of the question to use the word "impressive" to describe the SX6000, despite the fact that its performance isn't record breaking. It certainly warrants consideration by those who are in the market for a high-end ATA RAID card - the ATA RAID 2400A is limited to 4 channels and is somewhat more expensive, and some may think twice about purchasing an Escalade given 3ware's disappointing exit from the ATA RAID market. But, it ends up being the buyer's decision, as always. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Due to the consistently unrepresentative WinBench Diskmark scores we've been getting in our RAID reviews, we're no longer going to bother to comment on them in RAID articles. We'll continue to present them for the sake of completeness, but the only WinBench tests that we'll discuss are the sustained transfer rate, access time, and CPU utilization tests.
HOME | ARTICLES | LEADERBOARD | PERFORMANCE DATABASE | REFERENCE GUIDE
COMMUNITY | RELIABILITY SURVEY | SUPPORT SR! | ABOUT SR |
Copyright © 1998-2005 StorageReview.com, Inc. All rights reserved.
Write: Webmaster