企业和 IT 系统

IT Architect Resources

Architecture patterns for deploying MathWorks products

On-Premise Desktop Products

Architectural patterns for on-premise desktop deployment

MATLAB & Simulink (Individual and Designated Computer licenses)

Documentation

See Sections 1 & 4

MATLAB & Simulink (Network, Concurrent, and Campus-wide and Enterprise licenses)

Documentation

See Sections 3-6

On-Premise Server Products

Architectural patterns for on-premise server deployment

MATLAB Online Server

  1. User accesses MATLAB from a web browser
  2. Core services serve up web pages, authenticate users, assign MATLAB compute pods, enable licensing for the server
  3. A MATLAB container pod is provisioned for each user (which also checks out a license)
  4. User file storage and shared file systems are mounted

MATLAB Online Server Admin Guide

see Sections 1 & 2

MATLAB Parallel Server

  1. Users submits jobs to the cluster using their MATLAB client
  2. The headnode schedules jobs to run on the cluster
  3. MATLAB Parallel Server workers run user code on the cluster
  4. The user retrieves results back to their MATLAB client
  5. The job is executed across multiple nodes
  6. Monitor the job and retrieve results

Documentation

MATLAB Web App Server

  1. Users access the MATLAB Web Apps through a supported web browser
  2. MATLAB Web App Server license must be connected to a network license manager
  3. Authenticate user credentials for access control against an LDAP directory or an OpenID Connect identity provider

Quick Start Guide
Documentation

MATLAB Production Server

  1. MATLAB Production Server is stateless, and the load balancer should be set to route traffic via round robin or least connections schemes
  2. Multiple instances of MATLAB Production Server can be run for redundancy or improved performance. Each instance runs multiple worker processes that process requests
  3. All MATLAB Production Server instances must be connected to a network license manager

Quick Start Guide

Polyspace Bug Finder/Code Prover Server and Access

  1. Check in code to source repository
  2. Perform analysis on the code
  3. Host the results of the analysis
  4. Access the results via a web browser

Documentation

Cloud

Design patterns for cloud deployment.

MATLAB and Simulink

MATLAB Online Server

Any Cloud provider

MATLAB Online Server admin guide

See sections 1 and 2

MATLAB Parallel Server

Polyspace Server

Some products are available in several cloud delivery options. This table provides recommendations for specific use cases:

  Cloud Reference Architecture on GitHub Azure or AWS Marketplace Bring Your Own License (BYOL) MathWorks Cloud Center

Ability to customize, maximum flexibility

 

   

Move existing on-premises workloads to the cloud

 

 

 

Long-term trials, proof-of-concepts

 

 

 

Bursty workloads

 *

   

Run large jobs in clusters from your MATLAB desktop

   

 *

* MATLAB Parallel Server only

Learn about MathWorks Cloud products and services.

Containers and Orchestrators

Build or use prebuilt container images to deploy MATLAB and Simulink.

MATLAB dockerfile

MATLAB dockerfile

Use the dockerfile to build your custom MATLAB container image

MATLAB Production Server

Deploy MATLAB Production Server using the Helm Chart and prebuilt containers

  1. Client applications call MATLAB Production Server through RESTful calls or using the language specific client libraries
  2. The pre-built Docker container images for MATLAB Production Server and associated MATLAB Runtimes are pulled from containers.mathworks.com during installation
  3. If authorization is enabled, access is checked against an identity provider service
  4. The license server must be accessible from the Kubernetes cluster to provision licenses to the pods
  5. The deployable archives created in MATLAB Compiler SDK from your code are housed in Azure File Share, or NFS
  6. Run-time state can be store in REDIS
  7. Prometheus is configured to monitor the logs