Integration with timescale

[dberardo-com]

Is it possible to use mooncake on an existing table and in particular to query a materialized table managed by timescale?

If not, is it possible to create a mooncake table that is kept in sync with an underlying data table automatically?

[dpxcc]

It's not supported yet. For now, you need to manually keep it in sync using INSERT...SELECT from the source Timescale table to the destination Mooncake table

That said, this feature is on our roadmap, and we're actively working on it: #90

[dberardo-com]

Thanks for the reply

I might not have fully understood the feature request you shared, how is logical replication also covering the use case I proposed?

Btw, timescale has a nice feature called real time aggregates that basically tried to find all data points from the aggregates and the ones not found are going to be searched in the "raw" hypertable.

I am not requesting this note complex usage, but the possibility to query the aggregate table alone using a columnal store table would be nice.

So for now you are proposing to create a columnal store as a materialized view of the original table and then using it on an incremental way as needed. Right?

Does mooncake / duckdb also feature timeseries specific query function such as timescale's time bucket function?

Thanks

[dpxcc]

#90 aims to enable logical replication from an existing PostgreSQL table to a columnstore table and keep them in sync. I'm not familiar with Timescale, but I assume this could be used to create a columnstore mirror for a local or remote Timescale table

Does mooncake / duckdb also feature timeseries specific query function such as timescale's time bucket function?

DuckDB supports time_bucket: https://duckdb.org/docs/stable/sql/functions/timestamp.html#time_bucketbucket_width-timestamp-offset
We can expose it in pg_mooncake as well. Just opened a ticket to do so: #136

We’re working to improve pg_mooncake for time-series use cases and would love your input! Join our Slack channel so we can learn more about your needs and use cases @paurora17