Momentic, a developer of an AI powered software testing platform, has completed a rearchitecture of its caching system to support more than 2 million queries per day across 20 billion total entries. The company reported that the change has allowed it to maintain an average response latency of approximately 250 milliseconds.
The performance improvement was achieved by replacing PostgreSQL with ClickHouse, a column oriented database management system. The transition was led by Sergio De Simone, who detailed the technical shift in a recent company announcement.
Momentic operates a platform that uses artificial intelligence to automate software testing. The caching layer, which stores frequently accessed data to speed up retrieval, had become a bottleneck as the platform scaled. The original PostgreSQL based system could not keep up with the growing volume of daily queries and the total number of cached entries.
According to the company, the new architecture using ClickHouse delivers faster read performance for analytical workloads. ClickHouse is designed for high speed data ingestion and real time querying of large datasets, making it suitable for caching scenarios that involve heavy read operations. Momentic reported that the switch resulted in a significant reduction in query response times across the platform.
The migration involved reworking the data model and adjusting the application code to work with ClickHouse’s SQL dialect. The team also optimized indexing and partitioning strategies to maximize throughput. Momentic confirmed that the new system now handles the full traffic load without degradation during peak usage periods.
PostgreSQL remains a widely used relational database, but its row oriented storage format is less efficient for analytical queries that scan large numbers of rows. Column oriented databases like ClickHouse store data by column rather than by row, which can reduce I/O and improve query speed for aggregation and filtering operations. This design choice is increasingly common in high volume analytics and caching environments.
Momentic did not disclose specific cost savings or hardware changes associated with the migration. However, the company noted that the new setup reduced the number of nodes required to manage the caching layer, suggesting improved resource efficiency.
The shift reflects a broader trend among technology companies toward specialized database systems for distinct workloads. While PostgreSQL remains a strong choice for transactional applications, organizations handling large scale analytical queries are exploring alternatives such as ClickHouse, Apache Druid, and Google Bigtable.
Momentic plans to continue monitoring the performance of the new caching system and may extend the use of ClickHouse to other areas of its platform. The company has not announced a timeline for further migrations but said it expects the architecture to support future growth in both query volume and data size.
