Create a hypertable

Hypertables and chunks make storing and querying times-series data fast at petabyte scale.

TimescaleDB automatically partitions time-series data into chunks, or sub-tables, based on time and space. You can configure chunk size so that recent chunks fit in memory for faster queries.

A hypertable is an abstraction layer over chunks that hold the time-series data. Hypertables enable you to query and access data from all the chunks within the hypertable. You get all the benefits of automatic chunking for time-series data, alongside the simplicity of working with what looks like a standard, single PostgreSQL table.

Hypertables and chunks enable superior performance for shallow and wide queries, like those used in real-time monitoring. They are also good for deep and narrow queries, like those used in time-series analysis.

For more information, see .

1. Convert the regular table into a hypertable partitioned on the time column using the create_hypertable() function provided by TimescaleDB. You must provide the name of the table () and the column in that table that holds the timestamp data to use for partitioning (time):

2. Create an index to support efficient queries on the symbol and columns:

note

Because you often query the stock trade data by the company symbol, you should add an index for it. Include the time column because time-series data typically looks for data in a specific period of time.

TimescaleDB isn’t just for hypertables. Remember, TimescaleDB is PostgreSQL. When you have other relational data that enhances your time-series data, you can create regular PostgreSQL tables just as you would normally. For this dataset, there is one other table of data called company.

Creating regular PostgreSQL tables

1. Add a table to store the company name and symbol for the stock trade data:

To learn more about hypertables and best practices for configuring chunks, see Hypertable How-To. For information about how hypertables help with storing and querying data, see the .