Local
On one machine, such as an edge server next to the equipment or a laptop, tables live in a folder on disk and need no cloud account.
| Design | Data arrives as | Visible to queries after | Fits |
|---|---|---|---|
| A folder and cron | CSV or Parquet files in a folder | the next cron run (a minute) | gateways that write files; no code to write |
| Appending from Python | readings in your own program | each append (every few seconds) | an MQTT client, a Modbus poller, or any program that reads sensors |
| A local S3 | either of the above | the same | trying the cloud setups without an account |
Start with the folder and cron: it needs only the timelake command and a shell script. Move to Python when readings come from your own code rather than files. The cloud guides use the same pattern as the folder, with a bucket as the folder and a function as the script. Scheduled checks work with any of them.
Next
The same pattern on a cloud: AWS, Azure, Google Cloud, Cloudflare, Fly.io, Vercel, or Tigris. Quick start compares them.