The Speed Test tab in MemSQL Ops runs a simple performance benchmark that inserts randomly distributed data into MemSQL over 30 seconds, while measuring the throughput and analyzing the continually changing data set.

Running the Speed Test

To run the Speed Test, click the Speed Test tab in MemSQL Ops, and click Start Test to start the benchmark.

Once started, Ops will create a new dbspeedtest database and tbl table with the following schema:

CREATE TABLE tbl (
  id INT AUTO_INCREMENT PRIMARY KEY,
  val INT);

It then runs a series of INSERT statements as quickly as possible over 30 seconds that populate dbspeedtest.tbl with a random distribution of values for column val from 0-9.

As the test runs, you will see something similar to the screenshot below.

The left-hand side shows a Transactions view that displays the volume and rate of data inserted into MemSQL.

The right-hand side shows an Analytics view that displays the total volume of rows read, row scanning speed, and distribution of data across the 10 possible values in the val column. In real-time, you can also choose to query the database to see the data distribution for yourself by using the query shown in the web interface:

SELECT val, count(*) FROM dbspeedtest.tbl GROUP BY val;

The Speed Test page also shows:

• Speed Test Memory Usage: The amount of RAM consumed by data generated for the speed test
• Leaf Nodes: The number of leaf nodes in your cluster
• Aggregator Nodes: The number of aggregator nodes in your cluster

At the end of the Speed Test, an option will be provided for users to share their results with the MemSQL community via Twitter, as shown below:

Disabling the Speed Test

The Speed Test can be disabled in Settings > Config. There you will find the option to enable / disable the database Speed Test, which is enabled by default. Disabling the Speed Test prevents all users from running it. Running the Speed Test on a cluster in production is not recommended, as it may impact the performance of running database transactions.

Was this Article Helpful?

Thanks For Your Feedback!