Pricing

This is an estimate from public list prices, not a quote. Both sides store the same data for the same time on the same storage, and costs are per year once the retention window is full.

Your data

Each stream records count × frequency × record size for every hour it records, over 30-day months. Hot data is the hot share of that, kept for the hot retention. Cold data is the cold share, kept for the cold retention after the hot window. So stored hot data = recorded per month × hot share × hot days / 30, and the same for cold. Telemetry, metadata, and logs are stored smaller by the telemetry compression; binary data is stored as recorded. Both sides store exactly this data. Each month, 5% of the stored data is read back and leaves the cloud at its data out price.

ReductStore

ReductStore Pro costs $18 per TB per month of stored data, hot and cold, with a 1 TB minimum. It is a fixed price in each currency, never converted. Each instance runs on an m7i.xlarge on AWS, a D4s v5 on Azure, or a server priced like the AWS one on-prem. Telemetry faster than 1 Hz is written as one record per second, and ReductStore groups records into blocks of up to 64 MB before writing them to storage, so storage sees few, large objects.

Storage

Hot data is in the standard class (S3 Standard, Azure Hot). Cold data moves to the cheapest class whose minimum storage time fits the cold retention: S3 Standard-IA or Azure Cool from 30 days, S3 Glacier Instant Retrieval or Azure Cold from 90 days. Moving and reading cold data has its own request and retrieval prices, which are included. On-prem, hot and cold data both cost the price you enter per TB per month.

Foxglove

Foxglove Pro with 3 developer seats, one device per unit, and 20 query hours a month. Every stream is uploaded to Foxglove, which charges for indexing what is uploaded, for storing hot and cold data at the same price, and for bandwidth.

InfluxDB and TimescaleDB

Telemetry, metadata, and logs go to the database; binary data goes to the same object storage as one object per record. InfluxDB runs on 2 Amazon Timestream for InfluxDB 3 instances with data in object storage. TimescaleDB runs on Tiger Cloud with a primary and an HA replica, the newest 7 days on primary storage and older data on tiered storage. Tiger Cloud publishes only its entry compute price, so a 4 CPU service is estimated at 8 times that.

Engineering

Running ReductStore or Foxglove takes about 4 hours a month. A time-series database with object storage takes about 16: the upload pipeline, object naming, metadata schema, queries by time and label, and retention have to be built, which is spread over three years, and then maintained. Both are priced at the engineering rate.

Not included

Support, VAT, extra backups, and migration.

Prices

Cloud and vendor prices are in USD, for US East (N. Virginia) on AWS and East US on Azure. They vary by region, workload, and agreement.