Double Your SQL Server’s Performance Without Really Trying

How I Tested

Because it is difficult to control for variables in SQL Server (without a dedicated test lab) I choose to test Quickshift for SQL Server using Microsoft’s SQLIOStress Utility (downloadable free from Microsoft). This tool is designed to help identify potential I/O problems that could potentially cause SQL Server data corruption or loss. When this utility is run, it creates a SQL Server database and then performs extensive I/O testing on it, similar to what you might see in a production SQL Server (all without using an actual copy of SQL Server). The nice thing about this utility is that is produces a great stress on the I/O subsystem, and can do so consistently, from run to run. This helps to reduce the variability in testing. In addition, it provides a report of key performance indicators from each test run.

First, I ran the SQLIOStress Utility 10 times without Quickshift for SQL Server. Next, I installed the software, rebooted the server, and ran the utility 10 times using the identical SQLIOStress configuration. There was no other software running on this server. Here are some key results:

Before Installing Quickshift for SQL Server

Total Time to Run Test: 2 hours 33 minutes

After Installing Quickshift for SQL Server

Total Time to Run Test: 21 minutes

When running 10 passes of the test without Quickshift for SQL Server, it took 2 hours and 33 minutes to run. But once Quickshift for SQL Server was added (and nothing else changed), the identical test only took 21 minutes to run, an improvement of over 700% for this test.

In the Read Ahead simulation performed by the SQLIOStress Utility, Read Ahead speed was about 155 reads per second before using Quickshift for SQL Server, and about 1500 reads per second after, over a 900% increase in performance.

Average writes per second was about 2,548 before using Quickshift for SQL Server, and about 28,475 after, over an 1,100% increase.

Checkpoint writes per second was about 685 before Quickshift for SQL Server, and about 2,264 after, over an 300% increase.

These are just some of the results, but I think you can tell very quickly that Quickshift for SQL Server does boost SQL Server’s I/O performance. If you try this test yourself, you are bound to get different results, unless you have the exact same hardware and software configuration I was using.

While I have not personally been able to test the following, Quickshift claims that Quickshift for SQL Server has been Certified for Windows by VeriTest in these important areas: 1) Windows Fundamentals; 2) Installation and Removal; 3) Security Services; 4) User Interface Fundamentals; and most importantly for DBAs, 5) Cluster Services and Data Consistency.

What Kind of Performance Boost Will You Actually Get for Quickshift for SQL Server?

The testing I performed was artificial and you will most likely not see the kind of performance numbers you see here in the real world. According to testing performed by Quickshift, real world SQL Server performance boosts should be more in line of 200% to 300%, which is still very substantial. The amount of performance boost depends on a lot of variables, including:

  1. Your hardware (especially how much available RAM you have, the speed of your disk subsystem, and current CPU levels)
  2. Your application (how it is designed, coded, and used.)

The best way to find out for yourself how much of a boost Quickshift for SQL Server will give you is to try it. Quickshift offers a free trial of the software, so you can find out for yourself exactly what kind of performance boost you can expect.

Should You Purchase Quickshift for SQL Server?

If you are not satisfied with your SQL Server’s current performance, have already done everything you can do boost its performance, and if your hardware allows; purchasing Quickshift for SQL Server will provide a definite performance boost. But don’t take my word for it, try it yourself and see how it works in your unique situ


Leave a comment

Your email address will not be published.