How data modelling helps serve billions of queries in millisecond latency with real-time streaming

Users of financial data require their queries to return results with very low latency. As a financial data service provider, Bloomberg needs to consistently meet these requirements for our clients. HBase promises millisecond latency, auto-sharding, and an open schema. However, as HBase is a NoSQL database, support for transaction processing is not trivial. This talk will discuss a data modelling technique and use case where effective transaction processing can be achieved. We will also discuss how this data model helps us achieve real-time streaming, scalability, and millisecond read-write latency for billions of queries each day.