Self-hosted, distributed Durable Objects.
Cello runs a Cloudflare Workers application on your own machines:
Workers, Durable Objects, KV, Queues, D1, R2, Workflows, Cron Triggers,
and static assets, deployed from the wrangler.json you already have.
Each object is a cell: a named server with its own SQLite database.
Cello stores the long-term state in your S3-compatible, Google Cloud
Storage, or Azure Blob Storage bucket. It needs no serving control
plane or consensus service, and an inactive cell costs almost nothing.
📚 Docs: https://cello.ryuz.eu — Install · Operations · What Cello adds over upstream · Changelog
curl -fsSL https://cello.ryuz.eu/install.sh | shWorker projects deployed with cello deploy need
esbuild on PATH; asset-only projects do
not. Or run the image (Linux amd64/arm64):
docker run --rm ghcr.io/ryuzcorp/cello:latest --versioncello devOne node with a local object store — no Docker, no cloud bucket. The
Worker answers at http://127.0.0.1:9876, state lives in .cello/dev,
and the project rebuilds on every change without losing durable data.
cello deploy . --bucket s3://my-cells-bucket
cello \
--bucket s3://my-cells-bucket \
--listen 0.0.0.0:8080 \
--internal-listen 10.0.0.12:8081 \
--advertise 10.0.0.12:8081Start a second node against the same bucket and each write finishes as soon as the second node holds it on disk — much faster than the bucket round trip. No extra configuration: a node finds the other nodes through the bucket.
- Your Workers, unchanged. Deploy the Wrangler project you already have. The runtime, and every binding Cloudflare builds on Workers and Durable Objects, runs on your fleet.
- No control plane. A node is one
celloprocess per machine; the nodes that share one bucket are a fleet. A conditional bucket write gives a node ownership of a cell — no membership list, no leader election, no consensus service. Signed peer HTTP carries routing and replicated-log transport. - Durable by default. Cello does not answer a write until the data survives a failure (RPO=0): through the bucket, or through one or two peer nodes first when the fleet has them.
- Cheap at rest. An inactive cell is only objects in your bucket — one 8 GB node holds 1,000 resident cells, about $0.05 each per month.
- Hardened for other people's code. Operator token, tenant egress policy, project-confined builds, split deployment credentials, and non-dumpable serving processes. See Security.
- Observable. Traces and logs to your collector or to Parquet in the bucket, DuckDB-queryable. See Telemetry.
| service | example |
|---|---|
| Workers: fetch handlers, service bindings, JS RPC, Node.js compat | hello |
| Durable Objects: SQLite storage, alarms, hibernating WebSockets | counter |
| KV: list, metadata, expiration, bulk import | kv |
| Queues: producers, batching consumers, retries, dead letters | queues |
| D1: SQL databases, batches, migrations | d1 |
| R2: reads, writes, lists, multipart uploads | r2 |
| Workflows: durable steps, sleeps, events, pause and restart | workflow |
| Cron Triggers: one run for each occurrence across the fleet | cron |
Static assets: asset-only or with a Worker, _headers, _redirects |
static-assets |
| Dynamic Workers: runtime-loaded code and Tail Worker observability | dynamic-worker-tails |
| Containers (experimental): a Durable Object that supervises a container | container |
| Sandboxes: the Cloudflare Sandbox SDK on a container per sandbox | sandbox |
The Cloudflare compatibility page lists each gap in the services above.
cello diagnose --bucket s3://my-cells-bucket
cello cell list --bucket s3://my-cells-bucket
cello d1 migrations apply ledger --bucket s3://my-cells-bucketdiagnose probes every node lease; cell list prints one Class:ID
cell scope per line; d1, kv, and r2 run SQL and read and write
namespaces and objects through the fleet. The Operations
guide covers storage setup,
autoscaling, shutdown and rollout, upgrade notes, and every environment
variable.
A fleet runs one application, and every node loads its latest
deployment from deploy/current.json. Upgrades keep the bucket and the
node data directories: stop all old nodes, start the new binaries with
the same configuration. Mixed versions cannot share a serving fleet.
Cello is a renamed continuation of celld
by Deno Land Inc. — the What Cello adds over
upstream page documents every
improvement. To upgrade, swap the binary or image on the same bucket:
stop all celld nodes first (a mixed fleet is unsupported), rename
CELLD_* variables to CELLO_*, and start Cello. Cello reads
celld-written data and writes new data with cello names; see
Compatibility for the details.
Apache-2.0; see NOTICE.