InfluxDB 3 Core: a complete rewrite designed for speed and simplicity
InfluxDB has been a popular time series database for the better part of a decade, and the latest release represents years of work behind the scenes to address several major feature requests users have been asking for since the earliest days of the time series database.
Paul Dix, Co-founder and CTO of InfluxData, spoke at GrafanaCON 2025 about the recent release of InfluxDB 3 Core, the open source version of InfluxDB, which he described as the “culmination of a four and a half year journey of rebuilding InfluxDB.”
InfluxDB 3 Core is designed for simplicity and speed, Dix said, with new features for unlimited cardinality, SQL support, and the separation of compute and storage.
“We designed it from the ground up to be something that you could just have on your laptop in a few seconds or on a server,” Dix said. “It’s easy to get going. It’s a one-command-line install, native SQL, and InfluxQL out of the box.”
What is InfluxDB?
For those who aren’t familiar, InfluxDB is a time series database for metrics and events that can then be visualized in dashboards such as Grafana. It’s largely used on data that’s less than 24 hours old, and its designed to be highly performant for ingestion and querying. Use cases are geared toward server network monitoring, application performance, and sensor data of all kinds—from factories, powerplants, mining, water and waste, rockets, agriculture, satellites, vehicles, and more.

InfluxDB differs from other pure metrics databases because of the number of column types it supports, as well as its ability to do schema-on-write, Dix added.
“When you start up the database, you don’t have to create a database, you don’t have to create tables or anything like that. You start up the server, you start throwing data at it and it will create the logical database on the fly,” he said. “It will create the tables and create the columns. But once you’ve defined that, subsequent requests will validate against the schema that you’ve created.”
InfluxDB 3 Core: a rewrite to solve user challenges
The initial InfluxDB 3 release happened in 2023 with a hosted version, and the open source version (Core) was released this past April, completing a rewrite from Go to Rust.
“The question everybody has…is why do a big rewrite?” Dix said “And the truth is, we wanted to solve three big problems in the database that many, many people were asking us to solve.”
Unlimited cardinality
The first big problem they wanted to address was making sure high cardinality was no longer a blocker for users. People don’t want to have to think about the cardinality of the data they write to the database, Dix said. Metadata, which describes the time series, drives cardinality, but it also gives you greater precision with your observability. So they designed InfluxDB 3 to handle high cardinality without sacrificing on performance.
“If you want the ultimate precision, you’re going to have more of these dimensions, they’re going to be more unique values, and you’re going to have greater and greater cardinality in your observational data,” Dix said. “And we wanted a backend, a storage engine that would be able to handle this without driving up costs and being totally difficult to manage.”
SQL support
The next big change was adding SQL support. InfluxDB already had InfluxQL, which was similar but didn’t have full feature compatibility. “For people who are really familiar with SQL, it was different in sometimes frustrating ways.”
So InfluxDB 3 comes with native support for InfluxQL and SQL, making it easier for developers and data scientists to get started regardless of their experience with InfluxDB.
Object storage
Finally, InfluxDB 3 Core separates compute from storage. In previous versions, users had to have a locally attached SSD, which is OK for recent data, but not for longer-term data retention.
“People wanted to have the historical data available for query, but they didn’t want to pay for it on the most expensive storage medium available. They wanted to use object store,” Dix said.
To address this, they build a “diskless architecture” where you can specify the file system as the object store, which gives you multi-availability-zone durability at very little cost.
“One of the great things is we end up having stateless servers,” Dix said. “Now obviously the servers load up a bunch of state in RAM, but ultimately they’re stateless because you can boot up a new VM and just load all the state from object store. And this makes it operationally simple.”
Learn more about InfluxDB 3 Core
Ultimately, Dix said, InfluxDB 3 Core is built for simplicity and speed, with a one-command install, optimization for recent data queries, and an embedded Python processing engine, which is a big opportunity for engineers to push down advanced processing directly into the database.
To learn more about how a range of new features support this functionality—as well as some important info on limitations—check out the video from GrafanaCON 2025.
Grafana Cloud is the easiest way to get started with metrics, logs, traces, dashboards, and more. We have a generous forever-free tier and plans for every use case. Sign up for free now!