AT-1DB Cloud

Query your compressed archives in place.

Run SQL over compressed .at1 files in place, over HTTP — no decompress, no warehouse, no ingest. Your object storage (S3 / R2) holds the files; AT-1DB Cloud is the hosted query + verify layerover them. The engine runs server-side, so the browser never receives engine source, and every answer is checked against the file's SHA-256 manifest. Your storage, our hosted layer, nothing to run.

Your storage

The .at1 files live in your own S3, R2 or object store. AT-1DB Cloud queries them in place — the bytes never move into a warehouse, and there is nothing to ingest, copy or ETL.

SQL in place

Run predicates + projections straight against the compressed file over HTTP. Zone-maps and pushdown mean a selective query decodes only the blocks it touches — not the whole archive.

Verified every answer

The same query path re-checks the SHA-256 manifest, so every result is provably from the original, untampered file. Verification is free and account-less by design.

Nothing to run

The query + verify layer is hosted for you. No cluster, no engine to install, no server to keep patched — point it at a bucket and query. The compiled engine runs server-side only.

Storage stays yours; the query layer is managed

The split is the point. Cold, compressed .at1 archives sit in your bucket — you own the bytes, the region and the access policy. The hosted gateway (query_service/at1_query_server.py) holds the compiled engine and answers HTTP queries by fetching only the byte ranges a predicate needs. You get warehouse-style SQL over cold storage without paying to rehydrate it, and without standing up a cluster.

One API surface

A single POST /api/at1/query route, wrapped by the delinquency gate + metering. It forwards to the Python query gateway, which decodes only the blocks the predicate touches and re-verifies integrity on the way out.

# run SQL over a compressed .at1 in your bucket — over HTTP, in place
curl -X POST https://tinyfiles.io/api/at1/query \
  -H "authorization: Bearer $AT1_KEY" \
  -d '{ "file": "s3://acme-cold/txns_2026.at1",
        "sql": "select day, sum(amount) from t where region='EU' group by day" }'
#   -> { ok: true, rows: [...], bytes_read: 1.2MB, scanned: "0.4% of file", verified: true }

# the browser console (query explorer) speaks the same route
#   /dashboard/explorer

Who this is for

  • Cold-tier owners — years of compressed archives in S3/R2 that must stay queryable without a rehydrate bill.
  • Data platform teams — a query + verify layer over object storage, with nothing to operate.
  • Compliance & audit — every answer carries a re-checkable integrity guarantee against the original file.
  • Regulated / sovereign data — the bytes never leave your bucket or region; only queries cross the wire.
Honest scope: hosted service, deploy-on-demand

This is a hosted service, not a self-install. Turning it on is two environment variables — AT1_QUERY_URL and AT1_QUERY_TOKEN — pointed at the Python query gateway (query_service/at1_query_server.py) that holds the compiled engine and your files; no code change is required. Until that gateway URL is set, POST /api/at1/query returns { configured: false } and the query explorer runs in a clearly labeled demo mode. We don't claim a live endpoint before the deploy — it's ready to switch on for your account.

Your storage. Our hosted query + verify. Nothing to run.