Scale model deployments horizontally and vertically for optimal performance. You can also use model API routing to deploy simultaneous APIs for testing and production.
Scale horizontally for throughput-constrained model API endpoints. Typically these are endpoints that have many concurrent users. Consider horizontal scaling when downstream applications see long queues and running times from your endpoint.
By default, Domino schedules model instances to run on separate nodes when possible for availability and fault tolerance. However, you can set a model instance to run strictly on different nodes by setting the strictNodeAntiAffinity parameter to true in the API call when creating a new model API.
Model instances are also scheduled to run in separate availability zones when possible.
When you publish a model API, select the number of model API instances that you want to run at any given time. Domino automatically load-balances requests to the endpoint between these instances. A minimum of two instances (default) provides a high-availability setup. Domino supports up to 32 instances per model API.
|
Note
| Domino admins use the com.cerebro.domino.modelmanager.instances.defaultNumber Central Configuration key to change the default number of instances. |
Scale vertically for resource-constrained model API endpoints. Consider whether your endpoint requires complex tasks with more processing power. Scale model APIs vertically when downstream applications see long-running jobs for complex processes.
When you publish a model API, select a Resource quota that determines the amount of RAM and CPU/GPU resources available to each model API instance.
|
Tip
| The scaling settings are under Model APIs > <model name> > Settings > Deployment. |
|
Note
| If you make changes to scale the model API, you must restart it. |
You can set the degree of parallelism and scale all Python model APIs:
-
From the Admin screen, go to Advanced > Central Config.
-
Click the pencil icon for the
com.cerebro.domino.modelmanager.uWsgi.workerCountconfig key to update it.TipIf the com.cerebro.domino.modelmanager.uWsgi.workerCountconfig key is not listed on the Configuration Management page, click Add Record and enter the config key name at the end of the list in the key column. -
In the value column for
com.cerebro.domino.modelmanager.uWsgi.workerCount, set the value to greater than 1, which is the default. See the uWSGI project for more information. -
Click Save if you edit the key or Create if you add a new one.
-
Under the Configuration Management page title, the system displays: “Changes here do not take effect until services are restarted. Click here to restart services.” Click the here link and follow the directions to restart the services.
Domino supports basic and advanced routing modes to help you manage development and test deployments. To change routing modes, go to Settings > Deployment for each model API.
- Basic mode
-
In basic mode, one exposed endpoint always points to the latest successfully-deployed model API version. When you deploy a new version, the old version is shut down and replaced with the new one to maintain availability. Basic mode routes have the following signature:
Latest: /models/<modelId>/latest/model
- Advanced mode
In advanced mode, a promoted version and the latest version exist simultaneously. Advanced mode lets you point your clients to the promoted, production version, while giving you the ability to test with the latest version. When the latest version is ready for production, seamlessly switch it to the promoted version without downtime. Advanced mode routes have this signature:
Latest: /models/<modelId>/latest/model
Promoted: /models/<modelId>/labels/prod/model