QuestDB 8.0 with 50 percent more SQL performance and new VARCHAR data type

After just over a month, the developers of QuestDB are releasing version 8.0 of their time series database after version 7.4.2. Probably the most important new feature of the new major release is the new data type VARCHAR (variable character), which is intended to replace the previous data type STRING for future implementations. The developers felt that STRING, with its UTF-16 encoding and few optimizations, was too slow and unsuitable for modern applications. The new VARCHAR data type with UTF-8 encoding should require less space, and filter and sorting operations should also run significantly faster. The API has been adapted accordingly. A VARCHAR field can be up to 268 MByte in size, the entire VARCHAR column of a database is limited to 218 TByte. STRING will also work in newer QuestDB versions, but is marked as "deprecated" and will be removed at some point. The migration of a STRING column to VARCHAR should be easy:

ALTER TABLE
  example_table
ALTER COLUMN
  example_column TYPE VARCHAR;

Nevertheless, the developers encourage QuestDB users to read the additional information in the version 8.0 documentation.

ZFS compression now also for open-source version

As a special feature, the developers of QuestDB 8.0 point out that the free open-source version now also includes ZFS support and can use the compression there. Previously, only the enterprise version of QuestDB offered support for ZFS file systems and even required this. This change may have to do with errors that users have reported when using QuestDB on ZFS file systems. Previously it was also possible to set up a questdb pool under ZFS as compressed with -o feature@lz4_compress=enabled.

QuestDB is available in three versions: free as a restricted open-source product under the Apache 2.0 license, as QuestDB Cloud hosted by QuestDB at prices starting at around USD 272 per month and as a self-hosted Enterprise version, for which QuestDB does not specify any prices. The latter offers high availability, role-based access control, TLS for all protocols, data compression, cold storage and prioritized support. The Enterprise version is also used for the QuestDB Cloud.

What is a time series database?

QuestDB is a database specialized in time series (TSDB), comparable to InfluxDB, CrateBD, TimescaleDB or the almost antique RRDtool. All of these databases are optimized for storing and analysing time series such as sensor data, share prices and so on. A data set usually consists of a timestamp and one or more associated values (e.g. 16:02, 23 °C) as well as optional meta information. Time series databases are indexed via the time stamps, offer only weak consistency guarantees (values can be missing and are optionally interpolated) and are highly optimized for time-related queries. Older data can be automatically deleted or compressed. QuestDB explains this type of database clearly on its own website.

(mki)