InfluxDB has released an alpha build for testing its new time series indexing engine, which aims to support a large number of unique time series, with a goal of reaching 1 billion. The new engine moves the index to files on disk and uses techniques like Robin Hood Hashing for fast index lookups, allowing users to store millions of unique time series without significant memory impact. The feature addresses the issue of ephemeral time series, but does not yet solve limitations in scope and scale-out clustering. To enable the new engine, users need to download the alpha build and edit their config file with a new setting, and should only use it on testing infrastructure due to ongoing work and potential format changes.