postgresql pgbench Workload Simulation pg_bench - ghdrako/doc_snipets GitHub Wiki

pgbench is built in to PostgreSQL.

To start using pgBench, you first need to initialize a test database. This sets up a set of standard test tables with data for benchmarking:

pgbench -i -s 10 database_name

The -s 10 flag represents the scale factor, which determines the size of the test data. A higher scale factor creates larger tables, simulating more intensive workloads.

After initialization, you can run a benchmark test to evaluate how the database performs under load. For example:

pgbench -c 10 -t 1000 database_name

This command simulates 10 clients (-c 10), each executing 1,000 transactions (-t 1000), providing a realistic simulation of concurrent database activity. The tool then reports key performance metrics, such as:

Transactions per second (TPS): Indicates the throughput of the database.
Latency: Measures the response time for each transaction, helping identify potential delays.
Client-side and server-side performance: Allows you to evaluate both sides of the transaction process.

pgBench also supports advanced features like custom scripts, different transaction types, and adjustable workloads, making it highly customizable for various performance scenarios. Whether you're stress-testing your system or fine-tuning PostgreSQL for production workloads, pgBench provides valuable insights into how your database will perform under load.

my_benchmark.sh
#!/bin/bash

for run in {1..10}; do
  echo "select * from comments where blog_id = $(jot -r 1 1 1000);" >> queries.bench
done

echo "UPDATE comments SET view_count = 0 WHERE blogs.id = 1" >> queries.bench


# `-T/--time` time seconds
# `-j/--jobs` number of threads
# `-c/--client` number of clients
# `-M/--protocol` querymode = prepared
# `-r/--report-latencies`
#

pgbench --host localhost --port 5432 --username root --protocol prepared --time 60 --jobs 8 --client 8 --file queries.bench --report-latencies my_database_name"

chmod +X my_benchmark.sh
./my_benchmark

Another tool for PostgreSQL benchmarking is HammerDB. HammerDB may have a more realistic workload test environment out of the box.

Initialize pgbenchmark

./pgbench -i -p 5432 -d postgres

it creates pgbench_history, pgbench_tellers, pgbench_accounts, and pgbench_branches tables to run the transactions for benchmarking.

./pgbench -c 10                 # test with 10 clients with 10 tx per client
./pgbench -c 100 -T 300 -S -n   # ā€œ-Sā€ option - red only worload - secect , -n is to skip vacuuming on tables

pgbench typically runs transactions on its own tables. If you have a workload of 50% reads and 50% writes (or a 60:40 environment), you can create a script file with a set of statements to achieve the expected workload.

$ cat /tmp/bench.sql 
INSERT INTO test_bench VALUES(1,'test');
INSERT INTO test_bench VALUES(1,'test');
SELECT * FROM test_bench WHERE id=1;
SELECT * FROM test_bench WHERE id=2;

./pgbench -c 100 -T 300 -S -n -f /tmp/bench.sql

File the database test data

  • -i inicialize and fill data
  • -p port The database server's port number
  • -s scale_factor Multiply the number of rows generated by the scale factor. For example, -s 100 will create 10,000,000 rows in the pgbench_accounts table. Default is 1.
time pgbench -p 6000 -i -s 1000