What is data caching and how does Fabric use caching to improve performance?

• A. Data caching stores hot data locally or in a fast layer to accelerate repeat queries and notebooks, reducing compute and I/O overhead.
• B. Data caching stores all data permanently in a cloud object store for durability.
• C. Data caching eliminates the need for compute entirely in Fabric.
• D. Data caching only affects data archival and backup processes.

Answer: (A)

Data caching means keeping frequently accessed data in fast storage so it can be retrieved quickly instead of being read from slower sources every time. In Fabric, caching stores hot data locally or in a fast layer to speed up repeat queries and notebook runs. When you run analyses that touch the same data or produce the same results again, the system can serve those results from the cache, reducing the amount of compute and I/O work required. This leads to lower latency and faster iteration, especially for interactive work and iterative analytics.

Caching isn’t about keeping all data forever in a cloud store or eliminating compute. It’s a performance optimization: a temporary, fast-access layer that speeds up access to data that’s accessed often, with mechanisms to refresh or invalidate cached data when the underlying data changes. It complements durable storage, not replaces it, and it isn’t limited to archival or backup tasks.