What's the fastest way to get data into cockroachdb?

Source: Issue #4213

  1. For load testing, what’s the fastest way to get data into cockroachdb?

  2. When bulk loading an initial dataset, in the RDBMS world it is recommended one wait to add indexes until after the initial load, to make indexing faster (later) in batch. Is there a similar recommendation in force for cockroachdb? Any thoughts on batch loading vis-a-vis indexes?

  3. When loading bulk data into MySQL (say from a .csv comma separated file), there is a separate LOAD DATA INFILE facility to make this task easier. Is there comparable bulk loading functionality for cockroachdb?

I would find much use for:

  • a CSV loader
  • a loader for messagepack and json records
  • a loader for the time-series TMFRAME standard binary format; I wrote this recently for time series applications, see https://github.com/glycerine/tmframe There is go marshaling and unmarshaling code there already.

@vivek @mjibson