We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Unconference - Stream 3 - 15:45-16:45 - PGCon 2022
Discover the challenges and limitations of database benchmarks and learn how to create realistic and useful benchmarks for performance optimization, including the importance of data distribution, scalability testing, and foreign keys.
- Many benchmarks have been obsoleted or are no longer useful due to changes in software and hardware.
- Postgres and SQL Server have different procedural language implementations, making it difficult to directly compare them.
- HammerDB and TPCC are two useful benchmarks for OLTP workloads.
- PG bench is a complex tool with many configuration options, making it hard to use and requires expertise.
- Data distribution is crucial for benchmark results and should be carefully considered.
- Scalability testing can be challenging, especially when running multiple threads or nodes.
- Realistic benchmarks should test workloads similar to those experienced by actual customers.
- Foreign keys can significantly impact query performance and should be considered in benchmarking.
- It would be beneficial to have more peer-reviewed, open-source benchmarks that provide useful insights for performance optimization.