MATLAB Answers

Where can I find an overview of MATLAB Parallel Server?

11 views (last 30 days)
What parts encompass MATLAB Parallel Server and how do these parts work with one another?
Is there an overview to explain how MATLAB Parallel Server works?

Accepted Answer

MathWorks Support Team
MathWorks Support Team on 22 Jul 2021
Edited: MathWorks Support Team on 22 Jul 2021
The MATLAB Parallel Server environment is composed of three parts. The three parts are:
- MATLAB client installation including Parallel Computing Toolbox (PCT)
- MATLAB Parallel Server Worker nodes
- Job Manager/Scheduler (i.e. MATLAB Jobmanager, LSF, PBS, etc)
An overview:
The MATLAB Client:
-------------------------------------
Users interact with the cluster through a regular MATLAB installation which includes the Parallel Computing Toolbox (PCT). This is where tasks and jobs are created in MATLAB and sent to the cluster for processing. In order for this client to communicate with the Worker nodes of MDCS the Parallel Computing Toolbox (PCT) must be installed and properly licensed. The installation process for Parallel Computing Toolbox is the same as any other MATLAB toolbox.
MATLAB Parallel Server Worker Nodes:
--------------------------
The MATLAB Parallel Server Worker Nodes are the processing nodes of MATLAB Parallel Server, where tasks are received and completed. There is generally no user interaction with the worker nodes. Worker nodes must retrieve their licenses from a network license server. The quantity of keys in an MATLAB Parallel Server license refers to the maximum number of workers running at any time. If a worker node has multiple processors or cores you can start multiple MATLAB Parallel Server worker sessions on the node. The MATLAB Parallel Server installation on the worker nodes consists of all MATLAB & Simulink products.
The Job Manager/Scheduler:
-----------------------------------
The Jobmanager/Scheduler is the routing point for jobs, tasks, and results. The MATLAB client(s) communicate with the Jobmanager/Scheduler to submit jobs and tasks. The Jobmanager/Scheduler then relays tasks to available workers. Once a worker completes its task the results are returned to the Jobmanager/Scheduler, who in-turn relays the results back to the client session where they are received by the user.
For more information regarding MATLAB Parallel Server, please see the link below:
For a quickstart guide configuring MATLAB Parallel Server with MATLAB Job Scheduler
For an Application Guide for Application Cluster Administrators
For more information regarding PCT, please see the link below:
Detailed MATLAB Parallel Server installation instructions can be found here:
NOTE
: Starting in R2019a the following name changes occurred:
  • MATLAB Distributed Computing Server was renamed to MATLAB Parallel Server 
  • mdce_def was renamed to mjs_def
  • mdce binary was renamed to mjs

More Answers (0)

Tags

No tags entered yet.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by