Benchmarking Your MediaWiki Server
This article provides a guide to benchmarking your MediaWiki 1.40 server to understand its performance characteristics and identify potential bottlenecks. Regular benchmarking is crucial for maintaining a responsive wiki, especially as content and user base grow. We will cover various tools and metrics useful for assessing server performance. This guide assumes a basic understanding of server administration and command-line tools.
Why Benchmark?
Benchmarking helps you:
- Identify Bottlenecks: Determine if performance limitations stem from the CPU, memory, disk I/O, or database.
- Plan for Scalability: Understand how your server will perform under increasing load, allowing you to proactively upgrade hardware or optimize configurations.
- Measure the Impact of Changes: Evaluate the effectiveness of software updates, configuration tweaks, or caching strategies.
- Establish Baselines: Create a performance baseline to track degradation and identify issues before they impact users. See also: Server Performance Monitoring.
Tools for Benchmarking
Several tools can be used to benchmark a MediaWiki server. Here are some common options:
- ApacheBench (ab): A command-line tool for measuring the performance of HTTP web servers. Useful for simulating user load. See: Apache HTTP Server.
- wrk: A modern HTTP benchmarking tool designed for high-load testing. Often faster and more efficient than `ab`.
- MySQL Benchmark Tools: Tools like `mysqlslap` can benchmark the MySQL/MariaDB database server. See: Database Administration.
- vmstat: A system monitoring tool that reports virtual memory statistics, processes, CPU activity, and I/O.
- iostat: Reports CPU statistics and input/output statistics for devices and partitions.
- top/htop: Real-time process monitoring tools. See: Server Processes.
- PHPBench: A dedicated benchmarking tool for PHP code, useful for testing specific MediaWiki extensions or customizations. See: PHP Configuration.
Benchmarking the Web Server (Apache/Nginx)
We'll use `ab` to benchmark the web server's ability to handle requests. First, install `ab` if it isn’t already available. (On Debian/Ubuntu: `sudo apt-get install apache2-utils`).
A basic `ab` command looks like this:
ab -n 1000 -c 10 http://yourmediawiki.example.com/
* `-n 1000`: Sends 1000 requests.
* `-c 10`: Uses 10 concurrent requests. Adjust this based on your server's resources. Start low and increase carefully.
* `http://yourmediawiki.example.com/`: The URL of your MediaWiki installation.
Analyze the output of `ab`. Key metrics include:
* Requests per second: Indicates the server's throughput.
* Time per request: Average time to serve a request.
* Transfer rate: Data transferred per second.
Here's an example of how to interpret some results:
| Metric
|
Value
|
Interpretation
|
| Requests per second
|
150
|
The server can handle 150 requests per second.
|
| Time per request (ms)
|
66.67
|
Each request takes approximately 67 milliseconds to process.
|
| Transfer rate (KB/sec)
|
1200
|
Approximately 1.2 MB of data is transferred per second.
|
Benchmarking the Database Server (MySQL/MariaDB)
Database performance is critical for MediaWiki. `mysqlslap` is a useful tool.
mysqlslap --auto-generate-test --number-of-queries=1000 --concurrency=10 --engine=innodb --create-schema=mediawiki
* `--auto-generate-test`: Automatically generates test queries.
* `--number-of-queries=1000`: Runs 1000 queries.
* `--concurrency=10`: Uses 10 concurrent connections.
* `--engine=innodb`: Specifies the InnoDB storage engine.
* `--create-schema=mediawiki`: Creates a test schema named 'mediawiki'. WARNING: This will create a database. Use a test instance!
Key metrics to observe from `mysqlslap` include:
* Queries per second: Database throughput.
* Average latency: Average time to execute a query.
* Minimum/Maximum latency: Range of query execution times.
Here's an example:
| Metric
|
Value
|
Interpretation
|
| Queries per second
|
80
|
The database can process 80 queries per second.
|
| Average latency (ms)
|
12.5
|
Queries take an average of 12.5 milliseconds to execute.
|
| Minimum latency (ms)
|
1.0
|
Fastest query execution took 1 millisecond.
|
See also:
Database Query Optimization. Consider using a database connection pool like
PdoConnectionPool.
Benchmarking MediaWiki Specific Operations
Beyond the web and database servers, it's useful to benchmark specific MediaWiki operations, such as page viewing and editing.
You can use `ab` or `wrk` to simulate these:
*
Page Viewing: Benchmark accessing a typical article page.
*
Page Editing: Benchmark submitting a small edit to a page. (Requires authentication, so you'll need to include cookies in your `ab` or `wrk` requests).
*
Search: Benchmark performing a search query.
Here's a table summarizing typical server specifications for different wiki sizes:
| Wiki Size
|
CPU
|
Memory (RAM)
|
Disk Space
|
Database
|
| Small (Under 1000 pages)
|
2 cores
|
4 GB
|
50 GB
|
MySQL/MariaDB (Small Instance)
|
| Medium (10,000 - 50,000 pages)
|
4-8 cores
|
8-16 GB
|
250 GB
|
MySQL/MariaDB (Medium Instance)
|
| Large (Over 50,000 pages)
|
8+ cores
|
32+ GB
|
500 GB+
|
MySQL/MariaDB (Large Instance) / Database Cluster
|
Finally, consider monitoring resource usage during benchmarking with `vmstat` and `iostat` to identify specific bottlenecks. For example, high CPU usage suggests a CPU bottleneck, while high disk I/O indicates a disk bottleneck. See also:
Server Security.
Intel-Based Server Configurations
AMD-Based Server Configurations
| Configuration
|
Specifications
|
Benchmark
|
| Ryzen 5 3600 Server
|
64 GB RAM, 2x480 GB NVMe
|
CPU Benchmark: 17849
|
| Ryzen 7 7700 Server
|
64 GB DDR5 RAM, 2x1 TB NVMe
|
CPU Benchmark: 35224
|
| Ryzen 9 5950X Server
|
128 GB RAM, 2x4 TB NVMe
|
CPU Benchmark: 46045
|
| Ryzen 9 7950X Server
|
128 GB DDR5 ECC, 2x2 TB NVMe
|
CPU Benchmark: 63561
|
| EPYC 7502P Server (128GB/1TB)
|
128 GB RAM, 1 TB NVMe
|
CPU Benchmark: 48021
|
| EPYC 7502P Server (128GB/2TB)
|
128 GB RAM, 2 TB NVMe
|
CPU Benchmark: 48021
|
| EPYC 7502P Server (128GB/4TB)
|
128 GB RAM, 2x2 TB NVMe
|
CPU Benchmark: 48021
|
| EPYC 7502P Server (256GB/1TB)
|
256 GB RAM, 1 TB NVMe
|
CPU Benchmark: 48021
|
| EPYC 7502P Server (256GB/4TB)
|
256 GB RAM, 2x2 TB NVMe
|
CPU Benchmark: 48021
|
| EPYC 9454P Server
|
256 GB RAM, 2x2 TB NVMe
|
|
Order Your Dedicated Server
Configure and order your ideal server configuration
Need Assistance?
* Telegram:
@powervps Servers at a discounted price
⚠️
Note: All benchmark scores are approximate and may vary based on configuration. Server availability subject to stock. ⚠️