From the BigQuery page on partitioned tables:
Each table can have up to 2,000 partitions.
We planned to partition our data by day. Most of our queries will be date based but we have about 5 years historical data and plan to collect more each day from now. With only 2000 partitions: 2000/365 gives us about 5.5 years worth of data.
What is the best practice for tables wanting more than 2000 partitions?
- Create a different table per year and join tables when required?
- Is it possible to partition by week or month instead?
- Can that 2000 partition limit be increased if you ask support?
Update: Table limit is now 4000 partitions.