For the most part when someone mentions RAID (Redundant Array of Independent Disks), people tend to think of SCSI. Well awhile back, Promise Technologies came up with a cheaper alternative, IDE RAID. Improvements to IDE technology lead to the birth of the FastTrack66 card which utilized the new ATA/66 compatible hard drives to deliver even greater performance. It was inevitable that another company would create their own IDE RAID solution. Thus enter the new Iwill Super IDE RAID66. This new creation is Iwill's answer to the Promise FastTrack66 card. Both cards are designed to provide support for various RAID levels including, RAID 0, RAID 1, and RAID 0+1. The card can be used as a stand alone ATA/66 controller, but needs a minimum of two drives to utilize the RAID function. If you're confused about what exactly RAID is, let me give a brief summary.

Well, I've gone through this already in my previous review of the FastTrack66 controller, but it doesn't hurt to go over it again. RAID or Redundant Array of Independent Disks is basically a way to organize multiple hard drives into a single entity or array. So for example, if you have 10 hard drives, you can use RAID to treat them all as one drive. The way they are used together is determined by the level of RAID used. The main levels that the Iwill card supports are RAID 0, RAID 1, and RAID 0+1.

This level of RAID involves striping, where reads and writes of sectors of data are interleaved between multiple drives. Essentially, the workload is balanced between all the drives in the array, so performance is increased. It is recommended that you use identical hard drives for the greatest increase in performance. Although it is recommended, it is not necessary. The disk array capacity is equal to the number of drives times the smallest drive. The speed of the array will depend on the speed of the slowest hard drive. So for example, using a 10 GB and a 20 GB hard drive will create a 20 GB disk array. That's a lot of wasted space, so using two identical drives is definitely the way to go. Performance in RAID 0 is also determined by the stripe size used. The Promise FastTrack66 allowed you to specify a stripe size, which is the size chunks to interleave on the hard drives. Larger stripe sizes may be better for video editing and large file movement, while smaller stripe sizes might be better for everyday apps. Since the Iwill card does not let you specify the size, I assume it defaults to 16kb. A disadvantage of this level is that if one of the drives fail, the entire array is lost.

This level of RAID involves data mirroring, where duplicate data is written to a pair of drives while reads are performed in parallel. Fault tolerance is introduced here since each drive of the mirrored pair is installed on separate connectors. So, when one drive fails, the other one is still up with the data. You can rebuild the data of the damaged drive during reboot. Because both drives have the same data, the actual capacity will be half the total capacity of both drives. Again, if the drives differ in size, the larger drive will have wasted space. The workload is again distributed here to provide an increase in performance. However, it's not as high as the performance increase with RAID 0.

This level is basically a blend of the above two levels. It offers the same performance boost as RAID 0 while having the redundancy and mirroring capabilities of RAID 1. A minimum of 4 drives are needed to implement this since two pairs of drives are striped and those are mirrored on the remaining two drives. The major disadvantage of this is that it's expensive. To get the best performance you need 4 identical drives.

The card came with a user manual and two ATA66 cables. The cables were clearly labeled so you knew which end connects to the hard drive and which end connects to the motherboard. The manual is pretty straightforward. It describes pretty well the installation process and how to set everything up. However, it's not as detailed as Promise's manual. It's good though because that takes away some of the confusion.

The card itself is small. It's the smallest PCI card I have seen. The card has three IDE connections, two Internal and one external. I had some problems with the installation. For one thing, my case is rather large and the cables wouldn't reach the controller at all. I had to move around one of my drives to get it all working. Another problem is that one of the connections is on the bottom of the card. To me that makes it hard to have another PCI device in the next slot. Although it could probably fit, it will be a squeeze.

To install the controller, I had to reformat both my hard drives. I used two different hard drives for the testing since I didn't have two similar ones around. Other reviews out there use identical hard drives, but I believe that the scores for different hard drives gives a better sense of how the card performs. The major draw back of different drives is that if you use striping, the total size you get is double that of the smallest drive. So for someone who decides to buy a RAID card and wants to use to existing hard drives, this review will show how that would turn out.

The card uses the Highpoint Technologies chipset, which is the same used in my BP6. At first, I was afraid that there might be a conflict but it turned out nothing went wrong. One thing to note is that even though the drives were connected to the SIDE RAID66, the Highpoint BIOS on my BP6 would still register them as being there. Strange but I have seen no side effects. The rest of the installation went without any problems. Windows NT started up fine and I installed the drivers. The BIOS has already been flashed prior to installation for the best performance.

The test environment used here is identical to the environment used to test the FastTrack66. My previous testing was done using Windows 98, which didn't utilize the Iwill card to it's full potential. Comparing the cards tested in two different environments is like comparing apples to oranges. So I tested again using my usual benchmarking utilities, Winbench 99, Wintune 98, and CliBench MK II v1.0.0. What do these tests measure? They measure raw performance in different situations. Winbench tests the drive in various applications to provide a more general rating. My intent in using these benchmarks is to give variety so that you can measure performance in certain areas more clearly. All scores are in kb/s unless stated otherwise. The next section contains the updated scores.

The following are the benchmark scores using the RAID controllers.

These benchmarks are of the individual hard drives on the BP6 ATA/66 controller. These are basically the baseline scores.

Well, as you can see, the previous testing was definitely flawed. These scores are consistent with what I suspected all along. The old scores had the Iwill card crawling in comparison to the Promise card. Now it looks as if they are in fact relatively equal in terms of performance. The Promise card does perform better in some scores but clearly, the Iwill card is no slacker. The CliBench scores suggest that the transfer rate pretty much doubles in the RAID configurations. That is consistent with what is supposed to happen. It looks as if the Iwill card takes a bit more CPU time than the Promise card. The scores are very close, and it's hard to determine from these scores alone which is the better card.

Always keep in mind that two different hard drives are used, so the performance can still be increased. Undoubtedly, Windows NT offers a great increase in speed than Windows 98 does. But then again, RAID is usually used in servers which are run better with NT anyway. I guess NT is better equip to handle the file transfers, especially with NTFS.

There are numerous applications of the Iwill SIDE RAID66 controller. It's low price, $99 MSRP, is ideal for the budget server. Other SCSI RAID solutions cost upwards of $300 or more. Adding redundancy to your system is a great benefit. If you system has critical information, mirroring your drives can save you money in the long run in case something goes wrong. Small businesses can now afford to have their information safe without spending large amounts of cash. Recovering a damaged mirror is quick and simple, thus saving more time and money. Gamers can now experience the speed of RAID 0. The above scores show an increase in performance which will translate directly into faster game load times. No more hard drive thrashing when you get into a large Quake 3 FFA.

Professionals that use video editing tools frequently can now have good performance at a low price. Similarly, mp3 encoding and manipulation will be faster as well. Imagine crisp movie and sound playback while running background applications without any skipping. Basically anything that has to do with large or frequent data transfers will be see an increase in performance. This translates to most any application you can think off and improves over all system speed. In addition to being a RAID solution, the Iwill card is also a good ATA/66 card. You have the choice of using it as a regular expansion card and then later buying a second hard drive for RAID capability.

Well it had to happen, FastTrack66 vs SIDE RAID66. Which is better? Well that is a tough question to answer. Both have their strengths and weaknesses. The FastTrack66 is faster by a very small margin, however the Iwill card is also $50 cheaper. The Promise card has a great manual that covers everything about the card. The Iwill manual is compact, simple, and to the point. The Promise card has a better connector layout in my opinion, but the Iwill cad has a third connection for an external drive. With the performance on both cards being pretty much the same, the major determinant here is price and features. One thing the Iwill card doesn't support but the Promise card does is the ability to create a JBOD (Just a Bunch Of Drives) configuration, where the total capacity is equal to the sum of all hard drives. There is no performance gain in that configuration but just makes it convenient to treat many drives as one.

The Promise card also has the ability to choose the striping size, while you are stuck with default size with the Iwill card. It's a tough call. With the old scores, the Promise card clearly was the winner. The new scores tell a different story. I guess it depends on the application of the card. Iwill stresses that their card is very stable. My experience with both cards is that they are both stable even with my overclocked Celerons. There is no clear winner between the two, but I feel that the Iwill card with it's lower price beats out the Promise solution. You can find the Promise card for around $100 but why scour the net when Iwill is already at $99.

So what's the bottom line? I made a terrible mistake with the initial testing of the card. It appears that it performs at the same level as the Promise card, even better in some areas. What can I say, it's fast and affordable, even more so than the Promise card. You get high speed performance, data redundancy, and at a low price too. What more can you ask for? There are some minor flaws with the card but nothing warranting a reduction in rating. So my FINAL conclusion with this card is that it is the best IDE RAID solution for your money. I give it a score of 10/10.