Vendors can pay for more visibility on this website, for example a higher place in a ranking or more room on a page. Commercial relationships can also affect which products we feature and how prominently. Every vendor link is marked the same, whatever the arrangement behind it. The largest vendors in time series databases
cannot hold places 1 to 3. Places 1 to 3 go to challengers, vendors outside the best-known brands in a category,
wherever the category has them; the established names follow from place 4 in their ranked order. How it works: placement disclosure ·
editorial process.
The 14 tools, reviewed
#1 Tiger Data
Hourly-billed managed PostgreSQL for time series · United States · tigerdata.com
Published rates PostgreSQL Hourly billing
Tiger Data comes first because a buyer can read what it costs. Two named tiers list a starting compute rate and a storage rate per GB-month, and the 30-day trial needs no credit card, so an engineer can load real sensor or event data before anyone signs.
- Pricing: Usage-based. Performance compute starts at $30 per month and Scale at $36 per month, billed hourly, with storage at $0.177 and $0.212 per GB-month. Free plan: free trial.
- Best for: Teams that already run PostgreSQL
- Strengths: Hourly billing; Performance and Scale tiers; Point-in-time recovery up to 14 days (Scale).
Where it falls short
There is no ongoing free tier, only the trial, so a hobby project or a proof of concept that runs for months pays from day 31.
Wrong for
Teams that want a free-forever tier, or that need a purpose-built engine for market ticks or edge devices.
Visit Tiger Data →
#2 CrateDB
Distributed SQL database for time series and documents · Austria · cratedb.com
SQL Free plan Open source core
CrateDB earns second place for teams whose data is not only measurements. The same engine answers SQL over metrics, stores JSON documents and runs full-text search, which spares a shop from running a time series store beside a search cluster.
- Pricing: Usage-based. A free shared plan has 2 vCPUs, 2 GB RAM and 8 GB storage. Free plan: free plan.
- Best for: Mixed workloads, European data residency
- Strengths: Time series and SQL in one engine; JSON documents; Full-text search.
Where it falls short
The Dedicated plan has no free tier, and the Custom plan is priced on request. A general-purpose distributed engine asks for more cluster tuning than a narrow time series store, and the pages we read give no benchmark against specialist engines.
Wrong for
Buyers who only need cheap metric storage for a Prometheus setup, or who want the largest community of exporters.
Visit CrateDB →
#3 QuestDB
Open source time series SQL database · Not stated · questdb.com
Apache 2.0 SQL Parquet
QuestDB takes third place because its open source edition does the core job without a licence fee and speaks SQL with time-specific verbs. SAMPLE BY buckets data by interval and ASOF JOIN lines up two streams by timestamp, both common needs in trading and sensor work.
- Pricing: Open source + paid tiers. The open source edition costs nothing; Enterprise is sold on contact and no price is shown. Free plan: open source.
- Best for: Teams that want time series SQL, no lock-in
- Strengths: SAMPLE BY and ASOF JOIN in SQL; Tiered storage to Apache Parquet; Apache Iceberg tables.
Where it falls short
High availability, tiered storage and the lakehouse features sit in Enterprise, and its price is only on contact.
Wrong for
Teams that need a vendor-hosted database with published monthly rates and support from day one.
Visit QuestDB →
#4 InfluxDB
Time series database with usage-billed cloud · United States · influxdata.com
Usage billing Telegraf Open core
InfluxDB is the product most engineers name first, and its Telegraf collector, with more than 400 plugins, shortens the work of feeding it. The pricing page lists a rate for each meter on Cloud Serverless, so a buyer can model writes, queries, storage and egress.
- Pricing: Usage-based. Core is free. Free plan: free plan.
- Best for: Teams that want the default, widely supported choice
- Strengths: Cloud Serverless usage billing; Cloud Dedicated (single tenant); InfluxDB 3 Enterprise (self-managed).
Where it falls short
Core has no high availability and community support only.
Wrong for
Teams that need predictable flat monthly bills, or that want fully free high availability.
Visit InfluxDB →
#5 ClickHouse
Column-oriented analytical database, open source or cloud · United States · clickhouse.com
Columnar Open source Cloud tiers
ClickHouse fits when time series is one part of a larger analytical load. A team that stores billions of events and wants SQL over them can use the free open source server or ClickHouse Cloud, which runs on AWS, GCP and Azure and offers a $300 credit for 30 days.
- Pricing: Usage-based. The open source server is free. Free plan: free trial.
- Best for: Analytics teams with event and log volume
- Strengths: Basic, Scale and Enterprise tiers; Private networking (Scale); SAML SSO, CMEK, HIPAA and PCI (Enterprise).
Where it falls short
Cloud rates sit in a calculator that varies by provider and region, so a quick estimate is hard. Enterprise and bring-your-own-cloud plans are quote-based.
Wrong for
Small teams that just need to store and graph a few thousand metrics.
Visit ClickHouse →
#6 Grafana Mimir
Scalable long-term storage for Prometheus metrics · Not stated · grafana.com
AGPLv3 Prometheus Multi-tenant
Mimir solves the problem that Prometheus itself leaves open, keeping metrics for a long time and across many servers. It is horizontally scalable, multi-tenant and highly available, with data replicated and held in object storage such as S3, GCS or Azure Blob.
- Pricing: Open source + paid tiers. Mimir is free under AGPLv3. Free plan: open source.
- Best for: Platform teams running many Prometheus servers
- Strengths: Long-term Prometheus storage; Multi-tenant; Horizontally scalable and highly available.
Where it falls short
It is a metrics store only, so logs and sensor data from devices need another system. The AGPLv3 licence applies to self-hosted use, which some legal teams review.
Wrong for
Teams with a single Prometheus server or a small estate, or buyers who cannot accept AGPL terms.
Visit Grafana Mimir →
#7 VictoriaMetrics
Open source Prometheus-compatible metrics database · Not stated · victoriametrics.com
Apache 2.0 Prometheus Managed cloud
VictoriaMetrics is the practical pick for teams that already speak Prometheus and want a store behind it. The company publishes open source metrics, logs and traces projects, and the repository licence is Apache 2.0, which is permissive.
- Pricing: Open source + paid tiers. The open source core is free. Free plan: open source.
- Best for: Teams replacing Prometheus storage
- Strengths: Open source VictoriaMetrics, VictoriaLogs and VictoriaTraces; VictoriaMetrics Enterprise; VictoriaMetrics Cloud (fully managed).
Where it falls short
The pages we read show no prices for Enterprise, Cloud or Anomaly Detection, and no registered company location.
Wrong for
Buyers who need posted prices for a managed service, or an industrial historian with OPC support.
Visit VictoriaMetrics →
#8 TDengine
Industrial time series platform priced by tag · Not stated · tdengine.com
Tag pricing OPC and MQTT Industrial
TDengine is aimed at plants and field sites, and it connects to OPC, MQTT and Kafka, which is how machine data usually arrives. The pricing is unusual and clear: one number of tags, one price, whether self-hosted by annual or perpetual licence or bought as a cloud package.
- Pricing: Per tag / year. Up to 5,000 tags are free. Free plan: free plan.
- Best for: Factories and utilities with sensor data
- Strengths: OPC, MQTT and Kafka ingestion; JDBC and ODBC; Excel add-in.
Where it falls short
Standard Support is not included in the free tier, Premium Support is priced separately, and above 500,000 tags buyers must contact sales.
Wrong for
DevOps teams monitoring servers and containers, who think in series and hosts, not tags.
Visit TDengine →
#9 GreptimeDB
One database for metrics, logs and traces · United States · greptime.com
Apache 2.0 Unified data Managed plan
GreptimeDB is for teams tired of running one store for metrics, another for logs and a third for traces. It combines them in a single engine, speaks open protocols and is Apache-2.0 for the core.
- Pricing: Open source + paid tiers. The open source edition is free. Free plan: open source.
- Best for: Teams consolidating observability data
- Strengths: Metrics, logs and traces in one database; Open protocols; Enterprise Fully-Managed.
Where it falls short
It is the youngest product in this ranking, so references and third-party tooling are thinner.
Wrong for
Buyers who want a long track record in production or a large pool of engineers who already know the product.
Visit GreptimeDB →
#10 Prometheus
Open source monitoring system and time series store · Not stated · prometheus.io
Apache 2.0 PromQL Alerting
Prometheus is a standard for infrastructure metrics. It identifies series by metric name and key-value labels, queries them with PromQL and writes alerting rules in the same language, while Alertmanager routes the notifications. Everything ships under Apache 2, and the project is a CNCF graduated project.
- Pricing: Open source. Free under the Apache 2 licence; the cost is hardware and the people who run it. Free plan: open source.
- Best for: Teams monitoring cloud-native systems
- Strengths: PromQL query language; Pull-based scraping with service discovery; Alerting rules in PromQL.
Where it falls short
Each server stores data on local disk, so long-term retention and high availability are separate design problems that people solve with Mimir or VictoriaMetrics.
Wrong for
Teams that need years of retention, multi-tenant isolation or device telemetry from sensors.
Visit Prometheus →
#11 kdb+ (KDB-X)
Programmable time series engine for market data · Not stated · kx.com
Market data Quote only Vector search
kdb+ belongs on the list because KX describes it as the fastest programmable time-series engine, with a trading and surveillance audience in mind. The newer KDB-X adds native vector search, an MCP server and AI agent orchestration, with optional NVIDIA GPU acceleration.
- Pricing: Quote only. KDB-X is usage-based and monthly with a memory slider from 0 to 128 GiB and a Contact Sales button. Free plan: free trial.
- Best for: Trading firms and surveillance teams
- Strengths: KDB-X programmable time-series engine; Native vector search; MCP server.
Where it falls short
No prices appear on the pricing page, and the Market Data and Surveillance offerings are custom priced. The free-tier terms behind the trial link are not described.
Wrong for
Small and mid-sized teams with ordinary metric or sensor workloads and a fixed budget.
Visit kdb+ (KDB-X) →
#12 Apache IoTDB
Apache time series database for connected devices · United States · iotdb.apache.org
Apache 2.0 IoT Edge to cloud
IoTDB is built for devices rather than servers. Its edge-cloud architecture comes with a data synchronization tool, so readings gathered near machines can reach a central store. It aligns series from different sensors on time, compresses data heavily and connects to Hadoop, Spark, Flink and Grafana.
- Pricing: Open source. No price; the project uses the Apache 2.0 licence and the site names no commercial plan. Free plan: open source.
- Best for: IoT teams that want an Apache-governed project
- Strengths: Edge-cloud architecture; Data Synchronization Tool; One-click cloud installation.
Where it falls short
The pricing page link on the site returns a not-found page, and the licence is not restated on the home page.
Wrong for
Buyers who need a vendor to hold the pager or sign an SLA.
Visit Apache IoTDB →
#13 Apache Druid
Open source real-time analytics database · United States · druid.apache.org
Apache 2.0 Streaming Analytics
Druid targets dashboards that must answer within a second over data that is still arriving. It ingests streams without a separate connector, discovers schemas on its own, and tiers data so hot and cold storage can differ. A SQL API makes it approachable for analysts.
- Pricing: Open source. No price; the project uses the Apache 2.0 licence and the site lists no hosting. Free plan: open source.
- Best for: Teams building analytics on event streams
- Strengths: Interactive query engine with in-memory preloading; Stream ingestion without a separate connector; Tiering and quality of service.
Where it falls short
The site lists no deployment or hosting options, and operations and support stay with your team. A Druid cluster has several service types to size and monitor.
Wrong for
Teams that want a simple store for server metrics, or a vendor-run service.
Visit Apache Druid →
#14 OpenTSDB
Time series store on Hadoop and HBase · Not stated · opentsdb.net
Hadoop HBase Community
OpenTSDB suits an organisation that already operates Hadoop and HBase and wants a time series layer on the same cluster. It writes with millisecond precision and keeps raw data without downsampling, so nothing is thrown away. Support comes from a mailing list and IRC, and downloads sit on GitHub.
- Pricing: Open source. No price; open source, with no licence named and no hosted service on the site. Free plan: open source.
- Best for: Hadoop shops with existing HBase clusters
- Strengths: Runs on Hadoop and HBase; Millisecond-precision writes; Raw data kept without downsampling.
Where it falls short
The site states no licence, so a legal review has to look elsewhere. There is no hosted service or commercial support on the site.
Wrong for
Anyone starting without a Hadoop stack already in place.
Visit OpenTSDB →
Questions and answers
What is the best time series database in 2026?
Tiger Data ranks first of 14, best for teams that already run PostgreSQL. CrateDB is second (mixed workloads, European data residency) and QuestDB third (teams that want time series SQL, no lock-in). Places 1 to 3 go to challengers, vendors outside the best-known brands, wherever the category has them.
Do I need a time series database, or will PostgreSQL do?
PostgreSQL handles modest volumes well. Trouble starts when writes reach millions of rows per day and queries scan long time ranges, since a regular table then needs partitioning and manual retention.
Is Prometheus a time series database?
Yes, in a limited way. Prometheus stores time series and queries them with PromQL, but each server keeps its data on local disk, and the site describes it as a monitoring system.
How is InfluxDB Cloud billed?
Cloud Serverless bills by meter. The pricing page shows $0.0025 per MB written, $0.012 per 100 query executions, $0.002 per GB-hour of storage and $0.09 per GB transferred out, after a $250 starting credit.
What does TDengine charge for?
TDengine charges by tag count. Up to 5,000 tags are free forever on the self-hosted edition.
Can I run these databases on my own servers?
Yes for most. Prometheus, OpenTSDB, IoTDB, Druid, QuestDB, VictoriaMetrics, CrateDB, GreptimeDB, TDengine and Mimir publish open source cores, and InfluxDB offers a free Core edition. Tiger Data is a managed service, and kdb+ is sold by KX.
Which option suits industrial and IoT data?
TDengine and Apache IoTDB are built for it. TDengine connects to OPC, MQTT and Kafka and prices by tag, while IoTDB offers an edge-to-cloud design with a data synchronization tool and costs nothing under Apache 2.0.
Which products are best for market data?
kdb+ from KX is the specialist, described by KX as a programmable time-series engine, with KDB-X sold by usage and Market Data offerings by quote.
What is long-term storage for Prometheus?
A separate system that holds metrics beyond what one Prometheus server keeps locally. Grafana Mimir does this as a multi-tenant, horizontally scalable store with object storage on S3, GCS or Azure Blob, and VictoriaMetrics offers a Prometheus-compatible store.
Does CrateDB keep data in Europe?
CrateDB says it was founded in Austria and offers a managed cluster on STACKIT with data stored and processed exclusively in Europe, per its home page. Its paid plans also support AWS and Azure Private Link.
How do I move data out of a time series database?
Prefer products with open formats. QuestDB can hold older data as Apache Parquet and Iceberg tables, InfluxDB exports Parquet to data lakes, and SQL-based systems such as CrateDB and Tiger Data work with standard tools.