Keep one client's work off another client's machine
A rental is a whole host, so two projects are two machines rather than two processes. Stop one and its scratch disk goes with it.
Two clients on one machine is a conversation nobody wants to have, and one process boundary is a thin answer to it.
Why it is hard
Shared machines mean shared blast radius
One misbehaving job, one leaked path, one process reading a directory it should not, and the incident spans two engagements.
Cleanup is a promise, not a mechanism
Deleting a client's data at the end of a project usually means somebody remembering to, which is not the answer a client wants.
Per-project cost is a reconstruction
When everything runs on the same box, working out what a project actually cost means an afternoon with timestamps.
What answers it
A rental is a whole host
Not a slice and not a container on somebody else's box. Two projects are two machines, which is a boundary you can describe in a contract.
Ending it destroys it
The container and its scratch volume go when the rental does. Cleanup is what stopping means rather than a task somebody owes you.
Each project is its own line
Separate rentals are separate ledger entries with their own start, stop and rate, so what a project cost is read rather than reconstructed.
How billing works
per second, to the microeuro · no minimum · no egress fee · idle costs nothing
Metering starts when the container does and stops when it stops. The rate is frozen the moment you match, so a provider re-pricing cannot change a running rental, and an aborted run costs what it used.
What you use from Kracht
Related jobs
- ComputeTrain a model without buying hardwareTake an H100 for the length of a training run and give it back. No reservation, no minimum, and no card sitting idle between experiments.Read the job
- ComputeFine-tune on a scheduleA job that runs nightly needs a GPU for the hour it runs, not for the day around it. Per-second metering makes the difference the whole business case.Read the job
- ComputeServe inference when traffic spikesAdd capacity for the hours you need it and stop paying when the spike passes. Nothing continues to bill after the container stops.Read the job