Articles: Storage
 

Bookmark and Share

(3) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]

Database Patterns

In the Database pattern the drive is processing a stream of requests to read and write 8KB random-address data blocks. The ratio of read to write requests is changing from 0% to 100% with a step of 10% throughout the test while the request queue depth varies from 1 to 256.

You can click the following links to view the tabled results for IOMeter: Database:

We will build diagrams for request queue depths of 1, 16 and 256.

The SSDs split up into several groups here. The Colossus Lite, 120GB Agility and Solid 2 are contending for top place, the latter two models producing nearly identical graphs.

Next goes the Vertex 2 which is inferior to the leaders but also much better at writing than the smaller-capacity models. Its weak spot cannot be hidden, though. It cannot match the Indilinx-based products when it comes to pure reading.

The 60GB drives pass this test close to each other, their graphs nearly coinciding, whereas the 30GB models differ. The Onyx and the Vertex Turbo are roughly equals while the Solid 2 is slower, especially at about 10% writes.

When there is a request queue, the SSDs enable their NCQ algorithms. As they support but rather short queues, we have very similar pictures at queue depths of 16 and 256 requests. The Vertex 2 goes ahead, proving the superiority of its controller. It effectively reorders requests in the queue in such a way as to load as many memory access channels as possible at the same time. The Colossus Lite doesn’t look good at high percentages of reads where its performance is rather low. It is just as good as the other 120GB models at high percentages of writes, though.

The SSDs split up into groups based on their capacity again, the Solid 2 still being the slowest model in the lowest-capacity group.

Winding up this part of our tests, we want to show you diagrams that illustrate each SSD’s performance at five different request queue depths.

This group of SSDs all behave in the same way as they share the same controller and firmware. The differences are due to the different capacities as we have already found out. The smaller-capacity products are worse at writing and their graphs go lower in the right part of the diagram.

The graph of the Colossus Lite is indicative of its using an array of controllers. It does not benefit as much as its opponents from the request queue. We don’t know if it is due to some firmware deficiencies (the earlier firmware versions, e.g. 1.31, used to behave like that) or to the combining of the controllers into an array, but this SSD is not very good in its current form. However, we’ve got the junior model of the series and it is quite possible that large-capacity Colossus Lite drives deliver outstanding performance unachievable with SSDs based on one controller only.

Based on the SandForce SF-1200 controller, the OCZ Vertex 2 differs dramatically from its predecessors and behaves not unlike Intel SSDs. It accelerates at long request queue depths, which is indicative of its ability to correctly work with a longer queue than the other SSDs support. Take note that the graphs for requests queue depths of 4 and 16 differ a lot, although they coincide with the Indilinx-based models. It is also good that not only the speed of reading but also of writing grows up at longer queue depths.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]

Discussion

Comments currently: 3
Discussion started: 07/02/10 04:33:00 AM
Latest comment: 07/07/10 08:17:07 PM

View comments

Add your Comment