Glossary
Short definitions of the terms used on this site. Figures are measured on the production microVM runtime and explained on the methodology page.
- microVM
- A small virtual machine with its own kernel and hardware-level isolation that starts in milliseconds. Each MicroSaaS.app service runs in its own microVM.
- Scale to zero
- Suspending a service when no requests arrive and waking it on the next request, so an idle service uses a standby slot instead of a running one.
- Wake time
- How long a suspended service takes to answer after a request arrives. On the production microVM runtime the mean platform-reported wake time was 29 ms across 24 wakeups of a small test image.
- TTL (time to live)
- A deadline after which a service and its data are deleted automatically, and cannot be recovered. Stateful services have a time to live of at least 1 hour, set in whole-hour steps; the default is 1 hour. The maximum depends on the plan: 4 hours on Free, 24 hours on Developer and up to 30 days on Pro. The hosting platform supports an absolute scheduled delete and an idle-based delete; both were tested.
- Memory floor
- The smallest memory size at which a service booted and passed its own checks in our tests, rounded up to the next multiple of 32 MB. It shows how small a service can run and is not a sizing recommendation.
- Disposable service
- A service created for a short job, such as a test run or a prototype, that expires on a timer instead of being maintained.
- Control plane
- The component that receives API calls, creates microVMs, returns credentials and endpoints, and enforces TTLs.
- Gateway
- The hosting platform front door that terminates HTTPS and forwards traffic to an instance. It also limits request bodies to about 8 MB.
- Volume
- Persistent storage attached to one service at a time. Stateful services keep their data across stops and scale-to-zero cycles. Storage is included in your plan and allocated in 50 MB chunks. Each plan sets how many chunks are allocated to you.
- MCP (Model Context Protocol)
- An open protocol that lets AI agents call tools. MicroSaaS.app provides an MCP server, msaas-mcp, so agents can create and manage services.
Last updated