Continued from page 1 of 3ware Escalade 7500-4LP Review... 

 

IOMeter Benchmark Parameters 

This data will essentially be the same for both Windows 2000 Professional and Windows XP since they can be compared identically to each other.  In the beginning, I really wanted to compare results between Windows and Linux but there really isn’t a decent way of comparing a Windows OS to Linux since there are just too many resources that the Windows OS uses.  Not to mention that the IOMeter program is made for the Win32 Operating System and finding an equavilent program that will mimic the parameters is not out there to my knowledge.  There are quite a few I/O Benchmarks for Linux but we will not make use of them for this review.  For both flavors of Windows, I used the exact same partiton sizes of 20GB for the Windows install and the rest of the drive was left as unformatted and unpartitioned.  All of the programs were setup and configured exactly the same way across both installs.  As you can see in the table listed above, were the paramaters that were used for each Test Set and Run.  I didn’t have any problems with the program or the test system and a few hours later I had my results with IOMeter.  Below are the results of each Test Set and it’s individual Runs.

 Results of Test Set #1: Run’s 1 & 2 

The only variable that changed from each run within Test Set #1 was the Transfer Request Size of 8KB in Run #1 and an increased value of 64KB for Run #2.  The # of outstanding I/O's was kept at 64 for both Tests.  When I ran these tests, I was certain that Windows XP would have the edge over 2K Pro in performance although I really didn’t expect to see much of an increase of Total I/O’s.  I was mildly surprised to see that 2K Pro had the higher number of I/O’s over it’s successor.  For those of you that don’t know, the measure of Total I/O’s Per Second is an average number of requests implemented per second.  A request consists of positioning and read/write of the unit of the corresponding size.  So in plain English, the more Total I/O’s Per Second you have the faster your hard disks are.  Since we are using RAID, the results will be better than normal since we are basically taking one physical HDD and spreading it over two HDD’s.  You can imagine what testing would be like with all four HDD’s in this RAID array!

 

Result’s of Test Set #2: Run’s 1 & 2 

Once again, the older OS had more Total I/O’s than the newer OS.  The parameter’s here stayed the same from Test #1 which was the Transfer Request Size of 8KB in Run #1 and an increased value of 64KB for Run #2 but the # of outstanding I/O's increased from 64 to 128 for this round of tests.

 

Result’s of Test Set #3: Run’s 1 & 2 

Our third and final Test Set came with the expected outcome due to the increased # of outstanding I/O's up from 128 to 256 for the final results with the IOMeter Bench.  The settings used here would suggest that a server would be handling a very high disk load, essentially defragmentation of the disk(s).  Windows 2000 Professional consistently lead the Total number of I/O’s Per Second all through our tests.  I cannot explain this with any real confidence.  The only thing that I do know is that Windows XP is essentially built from the same code base as Windows 2000 and it stands to reason that Windows 2000 already has most of the major common problems and compatibility issuses worked out whereas XP is still a work in progress.  In any event, I was surprised by these results and have to pat the older OS on the kernel for a job well done.  These results would not have been comparable with any single hard disk system.  This can not only be contributed to the use of RAID and the sheer power of the onboard processor on the 7500 card itself.  This independent processor doesn’t steal many CPU cycles therefore leaving the main processor open to other duties.  Utilizing that 64-bit/66MHz PCI bus probably doesn’t hurt matters either.

Passmark Performance Test v4.0:

I have tested many programs for results such as what I have provided for you here today but I really like the benchmarking results that I get from this program called Passmark Performance Test v4.0.  Although this data is not a part of the IOMeter test suite I thought I would share with you some of the results that I received when trying it out while I had the Windows 2000 and XP installs up and running.  Out of all of the benchmarks inherent in this program, I will only make use of the Disk Test Suite to show the performance of the 7500 with our test system.  Below are some of the tests that will be run under the Disk Test Suite. 

This first test of this suite is called the Disk Sequential Cached Read test. A large test file is created for the disk under test.  The size of this test file is proportional to the amount of RAM available, which in turn stops the file from being cached in RAM.  The file is read sequentially from start to end.  The amount of data read in each individual read operation is always 16KB. 

The second test that is run is called the Disk Sequential Cached Write Test.  For this test, a large file is created for the disk under test.  The size of this test file is proportional to the amount of RAM available.  This stops the file from being cached in RAM.  The file is written sequentially from start to end.  The amount of data written in each individual operation is always 16KB. 

The third test that is run by the Disk Test Suite is known as Disk Random Cached Seek R+W.  A large file is created for the disk under test.  The size of this test file is proportional to the amount of RAM available.  This stops the file from being cached in RAM.  The file is then read randomly, a seek is performed to move the file pointer to a random position in the file.  A 16KB block is read or written then another seek is performed.  The amount of data actually transferred is highly dependant on the disk seek time. 

Here are the results of that testing for Windows 2000 Professional: 

Passmark Performance Test v4.0 Disk Suite Test
Disk Sequential Cached Read:	66.4MB transferred per second
Disk Sequential Cached Write:	76.3MB transferred per second
Disk Random Cached Seek R+W:	5.2MB transferred per second
Composite Results:	78.1 (Partial Result - Passmark)

 Here are the results of the same testing done with Windows XP: 

Passmark Performance Test v4.0 Disk Suite Test
Disk Sequential Cached Read:	62.6MB transferred per second
Disk Sequential Cached Write:	71.4MB transferred per second
Disk Random Cached Seek R+W:	4.8MB transferred per second
Composite Results:	74.9 (Partial Result - Passmark)

 With the same test conditions for both Windows 2000 Professional and Windows XP in another program (Passmark Performance Test v4.0) the results still show that Windows 2000 edges out Windows XP by a slim margin.  Although these tests were purely done out of fun, it has none-the-less shown me that my IOMeter results were not lying when it comes to the Windows 2000 OS performing better than Windows XP under my test conditions.  Although there are many other tests that can be run with this program, I was only interested in the Disk Performance Tests.  Like many other similar programs, you can download baseline files to get compariative results with your system.

3ware’s 3DM Software Management Utility:

I was quite pleased with the newest version of the 3DM Software that I downloaded from the 3ware site.  This interface is straightforward and easy to understand.  I only wish that the manual had more detail about the 3DM Software in there, maybe in a future revision or something.  I hope you didn’t scan through my warning about installing and using this newest version of 3DM as compared to an older firmware version on the controller?  Well, if you did, here is a quick link that will take you back there to re-read what I had to say about that, very important!  The interface is a web browser interface and gives you the ability to make some changes to your RAID array such as: Adding a drive, adding a spare (hot swapping), remove a drive, rebuild unit.  This all comes from the Configure page.  You can also schedule maintenance for your array and have the 3DM software automatically send you or your IT person an email if the utility detects something errorenous.  The Technical Detail View page displays an incredible amount of data for you to use at your discretion.  As you can see by the image below, there is more details there about your RAID array and its condition than you might not otherwise have known about without the 3DM Software installed.

 

Example of 3ware’s 3DM Software – Details Page

Final Thoughts and Conclusion:

The Escalade 7500-4LP ATA RAID controller is sitting on a mountain all by itself and there aren’t any contenders that can dethrone this mighty card.  The StorSwitch Architecture which basically applies network packet switching technology to the storage subsystem is the reason that this card works like no other.  The non-blocking switched fabric allows data to be simultaneously routed to multiple hard disk drives with the highest efficiency.  There is no doubt in my mind that this card can handle “any” amount of throughput that a small business server, gaming server, or the typical home user can throw at it.  The results of the I/O tests double what a single hard disk can handle in a similar test system minus the RAID controller.  Although the price is a bit high for most casual users, the person that is looking for a definite performance increase with a two to four hard disk system and an alternative to SCSI or the still emerging Serial ATA market can look into purchasing the Escalade 7500-4LP.  I have found this card ranging in price from $243.00 to $309.00.  Considering the alternatives, the price is still much less than that of a SCSI or even a Serial ATA solution and the performance, reliability, and functionality of the Escalade 7500-4LP is not far off the mark either.  I would use this product in any one of my home systems without question.

PCExtreme Rating:

I award the 3ware Escalade 7500-4LP ATA RAID controller a rock solid 4 out of 5 X’s

 

      

Warning: main(php/nav/footer.php) [function.main]: failed to open stream: No such file or directory in /home/pcx4010/public_html/escalade7500-4lp-2.php on line 450

Warning: main(php/nav/footer.php) [function.main]: failed to open stream: No such file or directory in /home/pcx4010/public_html/escalade7500-4lp-2.php on line 450

Warning: main() [function.include]: Failed opening 'php/nav/footer.php' for inclusion (include_path='.:/usr/lib/php:/usr/local/lib/php') in /home/pcx4010/public_html/escalade7500-4lp-2.php on line 450