Data Bench is a new proof-of-concept workload that can be used to measure the response-time latency of microservice transactions. This is a new type of workload that places the focus and analysis of computing environments on the handling, processing, and movement of data. Currently in Phase 1 proof-of-concept, this open-source implementation delivers all the required components for an out-of-the-box workload, in a format that is easy to download and easy to use. Data Bench provides a solid starting point for testing the interoperability and delivery mechanisms of future transaction benchmarks for microservices.
We introduce a new open-source workload: Data Bench. Data Bench is designed to help you tune, optimize, develop, and evaluate data-centric computing environments. This workload places the focus and analysis on the handling, processing and the movement of data. Specifically, Data Bench measures the response-time latency for two transactions using Kafka, Apache Spark*, and Apache Cassandra*. Note that other benchmarks use Twitter*, click-stream, or other unstructured data types for big data processing. Unlike those benchmarks, Data Bench uses contemporary online transaction processing (OLTP) structured data for the transactions and the data stored in Cassandra.
Download complete paper (PDF)