MySQL Performance Benchmarking
Compare MySQL database performance across Bare Metal Cloud instances.
Benchmarking Bare Metal Cloud
With 20+ available configurations, Bare Metal Cloud helps you optimize your workloads for maximum performance. To identify the best configurations for MySQL workloads, we tested out multiple instances using a simple Sysbench OLTP_RW test. The results are summarized below and should help you pick the right solution for your needs.
All instances leverage the new Intel Xeon Scalable processors (Cascade Lake Refresh), which are designed to provide performance boost for data-intensive workloads such as databases. Based on different thread counts and capacities, we compared 10 different benchmarks: Queries Performed (Read), Queries Performed (Write), Queries Performed (Other), Transaction per Sec, Queries per Sec, Latency (min), Latency (avg), Latency (max), Latency (95%), Latency (sum), Threads fairness (events), and Threads fairness (execution time).
The table below shows the three of those benchmarks that we considered the most relevant for these types of workloads. For the complete list, please contact [email protected].
MySQL Databases
For the purpose of comparing performance of different Bare Metal Cloud (BMC) instances, we tested a 100 GB MySQL database (8x tables with 50-million records, R/W/O of 70/20/10) on 16 Bare Metal Cloud configurations. Below is the summary of the results that should provide a reference for you when choosing your preferred instance.
Size: 100 GB
Ubuntu 18.04 OS
Database version 8.0.25.
Number of DB records: 50000000
70/30 read/write ratio
Benchmark Test Results
Instance | Threads | Transactions per Second | Queries per Second | Latency Average | Instance Price | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
s0.d1.small Single Intel E3-1240v3 (4 cores and 3.4 GHz) |
|
|
|
|
$0.10/h
$67.00/m |
||||||||||||
s0.d1.medium Single Intel E3-1270v3 (4 cores and 3.5 GHz) |
|
|
|
|
$0.12/h
$80.00/m |
||||||||||||
s1.c1.small Single Intel E-2276G (6 cores and 3.8 GHz) |
|
|
|
|
$0.18/h
$105.00/m |
||||||||||||
s1.c1.medium Single Intel E-2276G (6 cores and 3.8 GHz) |
|
|
|
|
$0.26/h
$164.00/m |
||||||||||||
s1.c2.large Single Intel E-2288G (8 cores and 3.7 GHz) |
|
|
|
|
$0.29/h
$184.00/m |
||||||||||||
d1.c1.small Dual Intel Silver 4210R (20 cores and 2.4 GHz) |
|
|
|
|
$0.44/h
$295.00/m |
||||||||||||
d1.c2.small Dual Intel Gold 5218R (40 cores and 2.1 GHz) |
|
|
|
|
$0.55/h
$378.00/m |
||||||||||||
d1.c3.small Dual Intel Gold 6230R (52 cores and 2.1 GHz) |
|
|
|
|
$0.61/h
$418.00/m |
||||||||||||
d1.c4.small Dual Intel Gold 6258R (56 cores and 2.7 GHz) |
|
|
|
|
$0.68/h
$475.00/m |
||||||||||||
d1.c1.medium Dual Intel Silver 4210R (20 cores and 2.4 GHz) |
|
|
|
|
$0.57/h
$393.00/m |
||||||||||||
d1.c3.medium Dual Intel Gold 6230R (52 cores and 2.1 GHz) |
|
|
|
|
$0.74/h
$516.00/m |
||||||||||||
d1.c4.medium Dual Intel Gold 6258R (56 cores and 2.7 GHz) |
|
|
|
|
$0.82/h
$573.00/m |
||||||||||||
d1.c1.large Dual Intel Silver 4210R (20 cores and 2.4 GHz) |
|
|
|
|
$0.77/h
$539.00/m |
||||||||||||
d1.c2.large Dual Intel Gold 5218R (40 cores and 2.1 GHz) |
|
|
|
|
$0.89/h
$622.00/m |
||||||||||||
d1.c3.large Dual Intel Gold 6230R (52 cores and 2.1 GHz) |
|
|
|
|
$0.94/h
$662.00/m |
||||||||||||
d1.c4.large Dual Intel Gold 6258R (56 cores and 2.7 GHz) |
|
|
|
|
$1.02/h
$719.00/m |
Test Methodology
To execute these tests, we used a simple Sysbench OLTP_RW test methodology that involved deploying the designated workload to each instance individually. Eight tables were used, each with 50 million rows (approx. 100GB of data). The load was incrementally increased with each test, starting with a single thread and scaling up to 8, 16, 32, 48, 64, 128, 254, and 512 threads.
Deploy BMC Servers Today