m-lab / etl-schema

All schema and views related to the etl pipeline and public bigquery tables.
Apache License 2.0
7 stars 2 forks source link

Risk of exceeding BQ partition limits #75

Open mattmathis opened 4 years ago

mattmathis commented 4 years ago

There is a 4000 partition limit for BQ tables, our unified views are already above 3950 days. It is not clear if the table partition limit applies to views.

Note also that our oldest web100 data is from 2009-02-18, which is more than 4100 days in the past: SELECT min(partition_date), date_diff(current_date(), min(partition_date), day) FROMmeasurement-lab.ndt.web100 (I have not yet explored the 50+ day discrepancy)

We need a long term plan going forward:

laiyi-ohlsen commented 4 years ago

We might have dodged a bullet?

mattmathis commented 3 years ago

This is a non-problem - legacy tables are not growing, and the new tables are column partitioned, which does not have this limit.