Glossary term
Glossary term
Infrastructure and Serving
Abbreviation for time to live.
Created for this library
An ML serving team configures a TTL on cached predictions so personalized scores refresh within an acceptable window.
A search team sets a TTL on cached query results so dynamic content refreshes within minutes.
A recommendation team uses a TTL on cached recommendations so personalization stays current without overwhelming the inference service.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License