For platform and data infra engineers
Keep the engine and the cluster you have. Add the worksheet, the per-query profile and the cost view your team keeps asking for.
You have clusters, jobs and notebooks. What you do not have is a worksheet, a profile for each query, or a way to see what a query cost. So engineers read job logs, and the bill is a single line item nobody can attribute.
One REST call, or the UI. XS is driver plus one executor.
Any Spark Connect client, including PySpark and notebooks. Existing DataFrame and Spark SQL code runs unchanged.
Every query gets an operator tree with duration and share of time, at a URL you can paste into a ticket.
No. The warehouse reads what your Spark deployment already reads. There is nothing to migrate to use the control plane.
No. The client protocol is stock Spark Connect. Your DataFrame and Spark SQL code is not rewritten.
No. It gives you warehouses on demand next to what you run. You can adopt it one workload at a time.
A stateless gateway, an on-demand driver and spot executors, in your own account. Suspend a warehouse and compute stops.
Early access opens in batches. One email when a batch opens, nothing else. Or run the whole stack locally tonight.
One email when signup opens. Stored with our mailing provider, never sold or shared.