Exporting results

NDJSON, CSV and Parquet, from the console, the API or the CLI.

Any query result can leave. There is no export fee and no rate limit beyond the usual query cost — data lock-in is not a business model we are running.

Format	Good for	Notes
`ndjson`	Re-ingesting elsewhere, streaming	Default. Streams, so no size cap
`csv`	Spreadsheets	Nested fields flattened to dot columns
`parquet`	Warehouses, notebooks	Columnar, typed, smallest on the wire

curl -X POST https://api.logstreem.com/v1/query/export \
  -H "Authorization: Bearer $LOGSTREEM_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "apl": "['\''api-gateway-prod'\'']\n| where _time > ago(7d) and status >= 500",
    "format": "parquet"
  }' \
  -o errors-7d.parquet

Export the aggregate, not the rows

Exporting 50 million raw events to count them in pandas is slower than summarize and a 40-row export. Push the work down to the query.

PreviousVirtual fields in the console NextDashboards