DataKitchen TestGen vs Percona pstress

DataKitchen TestGen profiles databases, generates data quality tests, and shows results and quality scores in dashboards. Its listed checks include 55 profiling characteristics, 32 hygiene detector tests, and 49 test types, including generated tests, business rules, and custom SQL. Monitors cover freshness, volume, schema, and custom metrics, using predictive models to flag changes such as late arrivals and schema drift. Supported databases include Snowflake, Databricks SQL, Azure Synapse Analytics, Azure SQL Database, SQL Server, OneLake, BigQuery, Redshift, Aurora PostgreSQL, Oracle 12c and later, SAP HANA, PostgreSQL, and Salesforce Data 360. It can also profile and test structured data in Iceberg, Parquet, Avro, ORC, CSV, and JSON through supported external-table options. TestGen runs in the customer’s infrastructure; DataKitchen says it does not host the application or receive customer data. It queries target databases with read-only access and stores results and metadata in its application database. The free Open Source plan is limited to one user, one connection, and one project. Enterprise costs $100 a month, billed per user and per database connection.

Percona pstress is a free, open-source Linux tool for concurrent database and crash-recovery testing. It creates randomized transactions from command-line options and runs concurrent workloads against a single server or cluster. Documented uses include checking database features, regressions, and recovery behavior for MySQL, Percona Server, and Percona XtraDB Cluster. A multithreaded C++ generator works with a Bash driver that coordinates concurrency and recovery steps. The driver can stop and restart a server between steps, change server variables, and continue with the preceding step’s data directory. Options control thread count, tables, records, transaction size, and SQL-operation probabilities. Trial directories normally retain the data directory, error logs, thread logs, mysqld binaries, and configuration. Building requires CMake 2.6 or later, a C++ compiler 4.7 or later, and development files for the target database version or fork. The repository identifies GPL-2.0 licensing. pstress runs self-hosted and does not verify that successful queries returned correct data.