Model-As-A-Service and managing them via Supervisor and Nginx

A “Model-as-a-Service” provides the capability to execute simulation models as a service. MaaS solely focuses on the application aspect of a model against data. There are two main usage patterns:-

1. The model can be pre-deployed, has a well-known service endpoint, and may be supported by supplemental data services. This is quite common for operational models used in a production environment.
2. The model can be dynamically deployed from the client before execution.
   Model service development for research purposes needs such a behavior.

Model-as-a-Service emerged as “a concept of being able to invoke re-usable, fine-grained software components across a network”

In blog, I mentioned, how it could be helpful in tackling parallelization of machine learning models along with consideration and avoidance of time taken to reload various models in parallel calls. Here in this post, we will see how to achieve MaaS and manager working with various models using Supervisor and Nginx.

RESTful services

MaaS involves web-services as the interface with which the client can communicate. Data is exchanged as structured text, such as XML or JSON (JavaScript Object Notation) in a specified syntax. However, data may also be passed to the service in the model’s native format.

There are several approaches to cloud computing in general, usually categorized by the service…