Vespa increases HTTP visiting throughput ten-fold or more

Vespa has introduced new features in release 7.499.15 to significantly enhance HTTP visiting throughput, addressing previous limitations in bulk read operations. The improvements include streaming HTTP responses to reduce latency and partitioning documents into independent subsets, allowing multiple visitors to access the full corpus concurrently across multiple containers or clients. This approach enables scalability, as throughput can now increase with the number of containers or clients, depending on the bottleneck. Benchmarks showed substantial throughput gains, such as an increase from 0.33Gbps to 1.12Gbps in a two-container setup and from 0.26Gbps to 3.3Gbps in a five-container setup, with slicing proving particularly effective in high-latency network environments. These enhancements ensure that Vespa users can perform large-scale document visits more efficiently by leveraging parallel processing and optimized data streaming.