TDengine Brings Open Source Time-Series Database to Kubernetes

TDengine today made available an update to its namesake open source time-series database so that it can now run on Kubernetes clusters.

TDengine CEO Jeff Tao says TDengine 3.0 is differentiated from other time-series databases in that it includes caching and streaming processing capabilities along with support for SQL and an ability to subscribe to data in a way that simplifies operations.

TDengine is also integrated with a range of analytics and observability tools including Grafana, Google Data Studio and Prometheus, Tao says. Features such as super tables, storage and compute separation, data partitioning by time interval and pre-computation make it easy to access data in a highly efficient manner, says Tao.

Stream processing with sliding windows combined with standard SQL syntax makes it possible to support both traditional continuous queries and event-driven stream computing, he notes.

Collectively, those capabilities make it possible to support high-cardinality applications involving billions of data points in a way that outperforms general purpose and legacy time-series databases when it comes to data ingestion, querying and data compression, he says. Overall, TDengine claims its database is two to five times faster than other time-series databases and provides 10x read/write performance improvements over a general-purpose database.

That capability will prove especially critical in edge computing scenarios involving, for example, internet of things (IoT) applications that generate terabytes of data from sensors and other types of data collectors in real time, notes Tao.

As Kubernetes continues to mature, the range of types of stateful applications being deployed on clusters continues to evolve. Time-series databases are now one of several database classes that can be deployed directly on a Kubernetes cluster. The challenge now is to find a way to deploy databases in a way that reduces both complexity and overall costs, notes Tao. Rather than simply providing a database, Tao says TDengine is including capabilities that IT teams would otherwise need to acquire, provision and support separately as part of its efforts to reduce the total cost of maintaining a time-series database.

It’s not clear what percentage of stateful applications will require a time-series database, but as more applications process data in near-real-time, the need to be able to apply timestamps to data is increasing. When an event occurred is just as critical to track as who was involved and what was transacted as more data is processed and analyzed where it is created and consumed. There are databases that support time-series data as part of a range of data types, but TDengine is making a case for a high-performance, purpose-built database designed to support application processing data in real-time at scale.

It will be up to each IT organization to decide how many different types of databases it will support. Developers, however, are now routinely building applications across a wide range of databases so database administrators (DBAs) that can optimize the performance of multiple types of databases are in high demand. Most of those DBAs, however, don’t have a lot of time-series database expertise, so as more applications process data in real-time the need to train DBAs to optimize these types of databases is likely to become a lot more acute.

Mike Vizard

Mike Vizard is a seasoned IT journalist with over 25 years of experience. He also contributed to IT Business Edge, Channel Insider, Baseline and a variety of other IT titles. Previously, Vizard was the editorial director for Ziff-Davis Enterprise as well as Editor-in-Chief for CRN and InfoWorld.

Mike Vizard has 1614 posts and counting. See all posts by Mike Vizard