Part of the Introduction to Streaming Data Concepts series

HSAP: Hybrid Serving and Analytical Processing

The term HSAP (Hybrid Serving and Analytical Processing) was coined as a variation on OLAP and OLTP to distinguish systems that are focused on providing high READ performance for analytical queries (complex columnar queries common to data warehouses) and lookup or “serving” queries typical to user-facing applications.

Origin

HSAP was originally coined by Alibaba in the marketing of their Hologres data product.

Approach

In order to serve both types of lookups equally well, data coming into HSAP systems generally has to be indexed in two different ways.

Examples of HSAP

Currently only Hologres follows this format.

Use Cases of HSAP

HSAP products are pitched as a simplification of the standard approach where data is duplicated into different services (like Snowflake for data warehouse, Postgres/MySQL read replicas, Redis for Application-serving, and new systems like Flink and Pinot for real-time.)

With HSAP, a business can theoretically replace multiple distributed services that are complex and expensive to manage, or create new capabilities around real-time analysis and user-facing analytics.