It is currently required to write code to load data into QuasarDB. This makes it difficult to evaluate QuasarDB, because: Datasets to import can be very large It is non-trivial to write optimised code to load this fast An official QuasarDB import tool that is able to understand common data formats (i.e. CSV) would solve this problem.
Using bulk insert is currently limited to only a single timeseries per operation. This creates friction in situations where the distribution of timeseries of incoming data is random: buffering strategies do not help, and you will end up with very few (< 5) data points per bulk insert operation.
A solution to this problem can be to allow insertions into multiple timeseries in a single operation.