Skip to content

Reduce size of uberstore indices #78

Open
@comcast-jonm

Description

An anonymous reviewer of an upcoming paper about Sirius suggests:

Design point: you don't need to keep an index of all records, just a sampling of them
(every 10th or 100th or 1000th); you'd use less RAM and once you seeked to the
closest preceding index entry in the log you'd read forward serially.

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions