Here at StorageReview.com, we're proud of the fact that we were the first hardware site to debut a database that allows readers to custom create their own comparisons between reviewed products. Our ability to do this, of course, hinges on our use of an unchanging testbed to test all drives. The amount of information we've been presenting, however, has been steadily rising over the past couple years.
The introduction of IOMeter test results into the StorageReview.com database combined with the existing "legacy" data already present (WinBench 99 on the old testbed/operating systems and ThreadMark 2.0) has created a mountain of figures that can drown even the most dedicated of SR readers. After all, unfamiliarity with IOMeter may very well be enough to push readers into frustration. As a result, we've attempted to streamline our Database input screen to create a more intuitive and manageable interface.
Due to some practical considerations, we've elected to separate the IOMeter database from the existing WinBench/ThreadMark collection. Our IOMeter trials throw two new variables into the mix that WinBench didn't worry itself about: the type of access pattern, and the depth of the load. By combining these settings with the different test results, we've come up with a scheme that allows the merging of two cases that in the past have been separated... comparisons and sorts.
Example #1- Single Drive Info: Care to see a straight listing of how your favorite drive (such as the Quantum Atlas 10k II) did in IOMeter? Leave the "Access Pattern" at its default ("All Access Patterns"), load at its default ("All Loads"), Test Results at its default ("All Test Results"), and select the Atlas 10k II in one of the six fields below, leaving all others blank. Proceeding with the "Let's Compare" button results in a compact sheet of information showing you how the drive fared in all measurements under our three environment settings and five load levels.
Example #2- Drive Comparison: Would you like to see the information conveyed in the single drive sheet as a comparison between two rivals (Atlas 10k II vs Seagate Cheetah 36LP)? Keep all the settings outlined in Example #1 and simply select the Cheetah 36LP in another drive field. The result will be tables displaying results of the drives (with the victor shaded) along with graphical representations of the performance difference.
Example #3- Drive Comparison, pruned: The results generated by Example #2 are admittedly a bit unwieldy in length. We can make the data much more readable by selecting a Test Result to further filter the data. Keep all the settings outlined in Example #1/#2 and choose "Total I/Os per second" in the "Test Results" field. As we mentioned before, the Total I/Os per second result is IOMeter's main report. MB/sec transferred and Average I/O Response Time are simply variants of the Total I/O score. The resulting page will present a comparison of the two drives restricted to I/Os per second only. This should make it easy for readers to determine which of two drives is "faster" through head-to-head comparison.
Example #4- Sorts of all Drives: The sort feature of our WinBench/ThreadMark database has proved quite popular. Sorts can also be performed on IOMeter data. For instance, let's assume you want to see the Total I/Os per second result under Heavy Load that all drives deliver in the Workstation Access Pattern. Simply select "Workstation Access Pattern," "Heavy" Load, and "Total I/Os per Second" in the Test Results drop-down menu. Leave the drive fields blank. The result will be a sort of all drives under the parameters selected.
The list of examples could go on for some time. Note that we've included a fourth drop down menu for the IOMeter database, a toggle to select between tables and graphs. The two presentations are tradeoffs- graphical representation of results make differences easier to grasp through a casual glance; graphs, however, take up more real-estate.
IOMeter is a very powerful and flexible benchmark. We've attempted to mirror such power and flexibility in our IOMeter database. We hope readers will find it quite useful.
Those readers comfortable with our existing database should be pleased to hear that we've kept the same basic interface for WinBench and ThreadMark (WB/TM) comparisons, effecting only slight changes to compact the forms. As has always been the case, the WB/TM database features a sort engine separate from the "drive comparator."
To compare WB/TM results between hard disks, simply select the units in question in the six-drive input field and hit the "Let's Compare!" button. The result will be a display of all WB/TM results associated with the selected drives, presented in a side-by-side comparison format. Not all drives have been tested in all OS/benchmarks; as a result, it's important to note that some fields will contain a "Not Tested" placeholder should one drive that's been tested in a given combination be compared to another that has not.
Sorts can be performed on all the results produced by WinBench 99, WinBench 98, and ThreadMark 2.0. To reduce the sprawling size of our previous input table, we've added drop-down menus next to each benchmark's selector to allow readers to choose the desired OS under which they wish to review results. Remember, not all drives are represented in all OS/Benchmark combinations, so all sorts will result in partial lists of our 92 reviewed drives.
This sums up the advice you may need to use the new Database. At the time of this writing, 21 drives have been tested in the new testbed. Plug away and form your thoughts; ours will come tomorrow!