Creating continuous aggregates

Now that you've been introduced to continuous aggregates, create your own continuous aggregate from your data.

Create an aggregate query to use in your continuous aggregate

The data used in this tutorial is second-by-second, or tick, data for stock trades. A popular aggregate pattern used for analyzing stock data is called a candlestick. Generally, candlestick charts use 4 different aggregations over a specific interval of time (for example, 1-minute, 5-minute, or 1-day aggregates):

  • high: highest stock price per interval
  • open: opening stock price per interval
  • close: closing stock price per interval
  • low: lowest stock price per interval

For this example query, the time_bucket() interval is 1 day. The high and low values can be found by using the PostgreSQL MAX() and MIN() functions. Finally, the open and close values can be found by using the first() and last() functions.

Creating an aggregate query

  1. Use a SELECT command to find the daily candlestick values for each stock in the entire 1-month dataset. This may take a few seconds to process all of the raw data into 1-day buckets:

    time_bucket('1 day', "time") AS day,
    max(price) AS high,
    first(price, time) AS open,
    last(price, time) AS close,
    min(price) AS low
    FROM stocks_real_time srt
    GROUP BY day, symbol
    ORDER BY day DESC, symbol;
  2. The results of the command look like this:

    day |symbol|high |open |close |low |
    2022-05-03 20:00:00.000 -0400|AAPL |164.9799| 159.32| 164.545| 159.25|
    2022-05-03 20:00:00.000 -0400|ABBV | 151.7| 150.99| 151.32| 147.59|
    2022-05-03 20:00:00.000 -0400|ABNB |158.7158| 148.84| 153.58| 145.88|
    2022-05-03 20:00:00.000 -0400|ABT | 115.2| 111.64| 115.08| 111.14|
    2022-05-03 20:00:00.000 -0400|ADBE | 421.93| 407.61| 419.53| 395.06|
    2022-05-03 20:00:00.000 -0400|AMAT | 118.47| 114.279| 117.95| 112.04|

Create a continuous aggregate from aggregate query

Now that you have the aggregation query, you can use it to create a continuous aggregate.

The CREATE MATERIALIZED VIEW command triggers the database to create a materialized view with the given name, in this case stock_candlestick_daily. In the next line, WITH (timescaledb.continuous) instructs TimescaleDB to create a continuous aggregate and not just a generic materialized view. Finally, the query from earlier is added after the AS keyword.

Creating a continuous aggregate from an aggregate query

  1. Using the aggregate query from the previous procedure, create a continuous aggregate for daily candlestick data:

    CREATE MATERIALIZED VIEW stock_candlestick_daily
    WITH (timescaledb.continuous) AS
    time_bucket('1 day', "time") AS day,
    max(price) AS high,
    first(price, time) AS open,
    last(price, time) AS close,
    min(price) AS low
    FROM stocks_real_time srt
    GROUP BY day, symbol;
  2. The query might take some time to run because it needs to perform these calculations across all of your stock trade data the first time. After the calculation results are stored, querying the data from the continuous aggregate is much faster.

The SELECT statement is the same query you wrote earlier, without the ORDER BY clause. By default, this code both creates the aggregate and materializes the aggregated data. That means the view is created and populated with the aggregate calculations from your existing hypertable data.

Run this query to get all the data in your continuous aggregate, and note how much faster this is than running the aggregate SELECT query on the raw hypertable data:

SELECT * FROM stock_candlestick_daily
ORDER BY day DESC, symbol;

Real-time continuous aggregates

By default, all continuous aggregates are created as real-time aggregates. This means that TimescaleDB will append (or UNION) recent data that has not yet been materialized through a refresh policy to the output of the continuous aggregate. In this diagram, that corresponds to the last three points of raw data, which belong to an incomplete bucket.


If you don't want real-time aggregation, you can disable it. Set the materialized_only parameter to true for your continuous aggregate. For more information, see the section on real-time aggregation.

To inspect details about a continuous aggregate, such as its configuration or the query used to define it, use the following informational view:

SELECT * FROM timescaledb_information.continuous_aggregates;


hypertable_schema|hypertable_name |view_schema|view_name |view_owner|materialized_only|compression_enabled|materialization_hypertable_schema|materialization_hypertable_name|view_definition |
public |stocks_real_time|public |candle_one_min|tsdbadmin |false |false |_timescaledb_internal |_materialized_hypertable_3 | SELECT time_bucket('00:01:00'::interval, stocks_real_time."time") AS bucket,¶ stocks_real_time.symbol,¶ first(stocks_real_time.price, stocks_real_time."time") AS open,¶ max(stocks_real_time.price) AS high,¶ min(stocks_real_time.price) AS low,|

Next steps

Now that your continuous aggregate is created, the next step is to create a continuous aggregate refresh policy.

Without an automatic refresh policy, your continuous aggregate won't materialize new data as it is inserted into the stocks_real_time hypertable. As mentioned before, when you query your continuous aggregate, TimescaleDB performs real-time aggregation to include any unmaterialized data. As the amount of unmaterialized data grows, this can slow down your queries.

With a continuous aggregate policy, your new data automatically materializes into your continuous aggregate, keeping the need for real-time computations low and your continuous aggregate queries efficient.

Learn more about continuous aggregates

See how real TimescaleDB users leverage continuous aggregates in the blog posts How FlightAware fuels flight prediction models for global travelers with TimescaleDB and Grafana and How I power a (successful) crypto trading bot with TimescaleDB.

Detailed information on continuous aggregates and real-time aggregation can be found in the continuous aggregates docs.

Found an issue on this page?

Report an issue!


Related Content