MySQL Benchmarking on m3.xlarge and m3.2xlarge Instances

In our last post about MySQL benchmarking, we benchmarked MySQL performance on an m3.large instance using both instance store and different types of Amazon Web Service’s (AWS) Elastic Block Store (EBS) storage.

In this post, we have extended the benchmarks on m3.xlarge and m3.2xlarge. We ran the benchmarks with and without optimizations. We explained the optimization and the rationale behind each of them in a previous post about using Sysbench for benchmarking.

Here is a comparison of the different metrics we measured in Sysbench:

Without Optimization

                                     With Optimization

Transactions Per Second

Transactions Per Second

Avg. Response Time

Avg. Response Time

95% Response Time


95% Response Time

 

We noticed a couple of interesting items in the graphs:

1. Even though we see an increase in the transactions per second (TPS) as we move to a more powerful instance type, we also see an increase in the response times. Ideally, the response time would decrease as TPS increases.

The reason behind this can be explained using Little’s Law, which states:

Occupancy = Latency x Throughput

As TPS increases, occupancy increases.

Throughput = Occupancy / Latency

The increase in occupancy leads to increased throughput amidst the increased latency. To better understand the relationship between latency and throughput, read this blog post about Little’s Law.

2. We can see that the  benefits of optimization is most evident in larger tables.

We also collected metrics for resource utilization. Here is a summary of resource utilization when the benchmark was run with and without optimizations. We saw a significant differences in disk octets when optimization was applied. We also saw differences in the way memory was consumed.


Without Optimization

With Optimization

CPU

CPU

Memory

Memory

Disk Octets

Disk Octets

Disk Operations

Disk Operations


There it is. Benchmarking results of MySQL on m3.xlarge and m3.2xlarge.

Did you find this useful?  

Interested in getting tips, best practices and commentary delivered regularly? Click the button below to sign up for our blog and set your topic and frequency preferences.

Sign Me Up!

 
 
 
September 23, 2014 / Benchmarking

About the Author

Flux7 Labs
Find me on: