Inside a resource, each deployment is a named copy of a Foundry model with its own TPM quota and deployment type. Requests identify the model by this deployment name.
Also called deployment.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Model deployment in context, with comparison tables and the common traps.
Terms in this definition
- TPM
Tokens per minute, the measure for Azure OpenAI quota and deployment rate limits. Going over it produces an HTTP 429 response.
Related terms
- actionOnUnmanage
Controls what a deployment stack does with resources that drop out of its template: detach them or delete them. The delete option takes over from complete deployment mode, which is on the way to deprecation.
- ALM
Short for application lifecycle management, the practice of promoting content from development to test to production. Fabric delivers it by pairing Git integration (the CI part) with deployment pipelines (the CD part).
- App Service Environment
Network-isolated, single-tenant deployment of App Service on the Isolated plan. Because its cost is much higher, it suits only workloads that demand isolation.
- Auto swap
An App Service deployment slot option that, after each deployment and once warm-up finishes, swaps that slot into production without manual action. Linux web apps and Web App for Containers can't use it.
- Autopatch groups
A way of grouping devices in Windows Autopatch, joining Microsoft Entra groups to update policies. Each includes a Test ring and a Last ring with room for up to 15 deployment rings, and one tenant can create 300 of them.
- Autopilot deployment profile
Controls how the out-of-box experience runs on Windows Autopilot devices: which deployment mode and join type to use, and whether every targeted device should be converted to Autopilot. Up to 350 of these Intune profiles can exist in a tenant.
- Autopilot diagnostics page
If the Enrollment Status Page profile allows it, pressing Ctrl+Shift+D during Windows 11 OOBE opens this page. It shows what the deployment is doing and can export logs, one of which is a CSV with the hardware hash.
- az ml online-deployment get-logs
Azure Machine Learning CLI command fetching container logs for a deployment, from the inference server by default or from storage-initializer, to troubleshoot issues like init() failures or missing packages.