Partition Forecast Value table

[peterdudfield]

Would be good to partition on forecast_value table as it getting to big

useful blogs

• https://stackoverflow.com/questions/61545680/postgresql-partition-and-sqlalchemy
• https://pypi.org/project/architect/
• https://github.com/sqlalchemy/sqlalchemy/issues/5313

[peterdudfield]

Currently there are '132,756,148' rows in the forecast value table. This is about

• 300 locations
• 288 runs in the day (each day_
• 0.5 yeras = 183
• 16 time horizons

= 150 million rows

Would be good to partittion per month, so that we have about 25 million in each table

[peterdudfield]

Good to also try clustering - 'CLUSTER forecast_value using ix_forecast_value_target_time' as this will also speed things up

[peterdudfield]

tried in #109

and upgraded in

• https://github.com/openclimatefix/nowcasting_forecast/pull/165
• https://github.com/openclimatefix/nowcasting_api/pull/154