Amazon Timestream

AWS’s time-series databases: a serverless engine now closed to new customers, and managed InfluxDB 2 and 3.

What it is: Time-series databases
Type: Managed service on AWS
Made by: Amazon Web Services
License: Proprietary service; Timestream for InfluxDB runs open-source InfluxDB 2.x (MIT) and InfluxDB 3 Core (MIT or Apache 2.0), while InfluxDB 3 Enterprise and 2.x read replicas add InfluxData licence fees
Facts checked: 6 October 2026

What it is

The name covers two different products. Timestream for LiveAnalytics, generally available since September 2020, is serverless: you create databases and tables and send records (a timestamp, dimensions such as well_id, and measures) through the AWS SDK. Writes land in an in-memory store replicated across three Availability Zones; after a retention period you set per table, data moves to a magnetic store laid out for large reads, and late-arriving data can be written there directly. Tables split into partitions by time and by dimension values, and SQL queries with time-series functions run on a worker fleet sized to each query, billed in Timestream Compute Units (one TCU is 4 vCPUs and 16 GB of memory). AWS closed LiveAnalytics to new customers on June 20, 2025; existing accounts keep using it, and AWS points newcomers to Timestream for InfluxDB.

Timestream for InfluxDB, generally available since March 2024, is managed InfluxDB priced like a database server: you pick an instance class and storage, and AWS runs the open-source engine in your VPC with patching, backups and optional Multi-AZ failover. It started with InfluxDB 2.7; read replica clusters need an InfluxData add-on bought in AWS Marketplace. Since October 2025 it also runs InfluxDB 3 as clusters that keep data as Parquet files in Amazon S3 and cache recent data on the nodes. The Core edition is one node without compaction, meant for recent data; Enterprise grows to 15 nodes (since March 2026), compacts files for long retention, and adds an InfluxData licence fee per vCPU-hour. In US East (N. Virginia) on 2026-10-06, a db.influx.large instance cost $0.239 an hour, LiveAnalytics queries $0.518 per TCU-hour, and InfluxDB 3 storage $0.023 per GB-month.

Why teams pick it

Existing LiveAnalytics users get a database with no servers to size or patch, where ingestion, storage and queries scale separately and data tiers from memory to magnetic storage on its own
Timestream for InfluxDB keeps the open InfluxDB APIs, line protocol, Telegraf and client libraries, so code moves between self-hosted InfluxDB and AWS
InfluxDB 3 clusters keep data in S3 and bill storage on what is actually stored, separately from compute
AWS operations built in: Multi-AZ standby with automatic failover, backups and restore, maintenance windows, customer managed KMS keys, VPC and IAM integration

Watch out for

LiveAnalytics is closed to new customers since June 20, 2025; AWS’s migration guide exports data to S3 and loads it into Timestream for InfluxDB or Aurora/RDS PostgreSQL
Timestream for InfluxDB is not serverless: instances bill every hour they run (per second, 10-minute minimum), and Multi-AZ doubles both instance and storage prices
Licence fees on top of AWS charges: 2.x read replicas and InfluxDB 3 Enterprise are billed by InfluxData through AWS Marketplace, and the read replica add-on can only be subscribed to in the console
InfluxDB 3 Core is a single node without compaction that AWS doesn’t recommend for long-term storage, and every InfluxDB 3 cluster pays for at least 200 GB of storage
No host access, and InfluxDB 2.x instances top out at 16 TiB of storage; AWS advises staying under 10 million series per instance

Alternatives

What all time-series databases have in common →

Apache IoTDBOpen sourceAn Apache time-series database for industrial IoT that stores everything in its own open file format, TsFile, and lets you model devices as a tree of paths or as SQL tables.+1 TimechoDB ArcOpen sourceA young single-binary time-series database in Go that writes incoming data as Parquet files and queries them with DuckDB SQL.

ArcticDBCommercialMan Group’s serverless DataFrame database: a Python library with a C++ engine that stores versioned pandas DataFrames directly in S3, Azure Blob Storage or local LMDB.

GreptimeDBOpen sourceA Rust database that stores metrics, logs and traces as tables on object storage, queried with SQL or PromQL.

InfluxDBOpen sourceThe best-known time-series database, fed by line protocol and Telegraf; now in its third generation, a Rust rewrite that keeps data as Parquet files on object storage.

kdb+CommercialA columnar time-series database that is also an interpreter for q, an array language, used for market data across trading and risk.

QuestDBOpen sourceA columnar time-series database in Java, C++ and Rust that ingests millions of rows a second and answers time-series SQL, with old partitions kept as Parquet.

TDengineOpen sourceA clustered time-series database for IoT and industrial data that gives every device its own table, grouped under “supertables” with tags.

TimescaleDBOpen sourceA PostgreSQL extension that turns ordinary tables into time-partitioned, compressed time-series tables.

Workers Analytics EngineCommercialCloudflare’s time-series analytics store for Workers: a non-blocking `writeDataPoint()` call to record events with unlimited cardinality, adaptive sampling to keep it cheap, and a SQL API to query them.

How it fits with other technologies

What Amazon Timestream uses

What uses Amazon Timestream

InfluxDBis offered as a managed AWS service byAmazon TimestreamTime-series databases

Create a Multi-AZ InfluxDB 2 instance and write line protocol · bash

ID=$(aws timestream-influxdb create-db-instance \
  --name plant-metrics \
  --db-instance-type db.influx.large \
  --db-storage-type InfluxIOIncludedT1 \
  --allocated-storage 100 \
  --deployment-type WITH_MULTIAZ_STANDBY \
  --username admin --password "$INFLUX_ADMIN_PASSWORD" \
  --organization plant --bucket sensors \
  --vpc-subnet-ids subnet-0a1b2c3d subnet-4e5f6a7b \
  --vpc-security-group-ids sg-0123abcd \
  --no-publicly-accessible \
  --query id --output text)

# wait for AVAILABLE, then read the endpoint
aws timestream-influxdb get-db-instance --identifier "$ID" --query '[status, endpoint]'

# from inside the VPC: write with the InfluxDB v2 API on port 8086
curl -X POST "https://$ENDPOINT:8086/api/v2/write?org=plant&bucket=sensors&precision=s" \
  -H "Authorization: Token $INFLUX_TOKEN" \
  --data-binary 'pressure,well=w17 psi=812.5 1791273600'

LiveAnalytics, InfluxDB 2 or InfluxDB 3?#

Three engines share the Timestream name. Only the two InfluxDB options are open to new AWS accounts.

	LiveAnalytics	Timestream for InfluxDB 2	Timestream for InfluxDB 3
New customers	No, closed since June 20, 2025	Yes	Yes
How it runs	Serverless; AWS sizes everything	One instance you size, optional Multi-AZ standby or read replica cluster	A cluster of nodes you size: Core is one node, Enterprise up to 15
Where data lives	Memory store, then magnetic store, by table retention settings	Instance storage (Influx IOPS Included SSD, up to 16 TiB)	Parquet files in Amazon S3, cached on the nodes
Queries	SQL with time-series functions	Flux and InfluxQL	SQL through the InfluxDB 3 APIs
You pay for	Data written, memory and magnetic storage, query TCU-hours	Instance-hours and provisioned storage	Node-hours, stored data (200 GB minimum), plus an InfluxData licence per vCPU-hour for Enterprise

List prices in US East (N. Virginia)#

On-demand list prices checked on 2026-10-06 from the AWS price list. Data transfer out and backups are extra.

Item	Price
LiveAnalytics: writes	$0.50 per GB ingested; each write is counted in 1 KiB units
LiveAnalytics: memory store	$0.036 per GB-hour
LiveAnalytics: magnetic store	$0.03 per GB-month
LiveAnalytics: queries	$0.518 per TCU-hour, per-second billing with a 30-second minimum
InfluxDB 2: `db.influx.medium` (1 vCPU, 8 GiB)	$0.120 an hour, $0.239 with Multi-AZ
InfluxDB 2: `db.influx.large` (2 vCPU, 16 GiB)	$0.239 an hour, $0.478 with Multi-AZ
InfluxDB 2: storage, 3K IOPS tier	$0.10 per GB-month, $0.20 with Multi-AZ
InfluxDB 3: `db.influxIOIncluded.large` node (2 vCPU, 16 GiB)	$0.264 an hour per node, S3 requests included
InfluxDB 3: object storage	$0.023 per GB-month, minimum charge for 200 GB

InfluxDB 3 Enterprise and InfluxDB 2 read replicas add InfluxData licence fees, billed separately through AWS Marketplace.

See it in a real stack

No architecture uses Amazon Timestream yet. Drop it into the playground and wire it up yourself.Open playground →

Benchmarks

Benchmark · Tiger Data (formerly Timescale)Time Series Benchmark Suite (TSBS)A Go toolkit that generates synthetic server-monitoring or IoT time series, bulk-loads it into a database and times a fixed set of time-series queries.Includes Amazon Timestream · 6 caveats to read first

How to read a benchmark →

Where it came from2018–2025

AWS announced Amazon Timestream at re:Invent in November 2018 as a preview of a serverless time-series database for server logs, sensor data and industrial telemetry. Its pitch was cost and scale: AWS claimed it would process trillions of events a day at one tenth the cost of relational databases, keeping recent data in memory and moving history to cheaper storage automatically.

General availability took nearly two years, until September 2020. Scheduled queries, multi-measure records and direct writes to the magnetic store followed in 2021, and compute-unit query pricing in 2024. Meanwhile many teams already ran InfluxDB, so in March 2024 AWS added a second engine under the same name: managed open-source InfluxDB, in the style of RDS.

In May 2025 AWS said it would close the original engine, renamed Timestream for LiveAnalytics, to new customers from June 20, 2025, and recommended InfluxDB instead. Managed InfluxDB 3, with data on S3, arrived in October 2025.

Why it’s built this way

Serverless, with ingestion, storage and queries decoupled: AWS wanted each layer to scale on its own and to avoid correlated failures as the system grows, so customers never size servers.
A memory store and a magnetic store in one table: Most time-series writes and dashboard reads touch recent data, so it stays in memory, while older data moves to storage organized for large scans, and one query reads across both.
Partitions split by dimension values as well as time: Nearly all writes arrive around the current time, so splitting by time alone would not spread write traffic or let queries skip data.
Run open-source InfluxDB for the second engine: Teams with InfluxDB code, Telegraf agents and dashboards could move to a managed service without rewriting against a new API.

Timeline

2018
Timestream announced in preview at re:Invent
2020
Generally available in September in four Regions
2021
Scheduled queries, multi-measure records and magnetic store writes
2024
Timestream for InfluxDB launches in March; TCU query pricing in April
2025
LiveAnalytics closed to new customers in June; InfluxDB 3 in October

Sources18 · facts checked 6 October 2026

Improve this page on GitHub