Can someone explain the difference between Parallel Server and Production Server

New to MATLAB but working to standup a horizontally scalable implemenation in AWS for a client. All indications are that this is achieved using Parrallel Server or Production Server. Architectures (based on the docs and Cloudformation template analysis) appear to be very similar i.e. head/console instance AMIs and multiple worker instances managed via an autoscale group. The Production Server product apparently allows for an optional Redis persistence implementation but it's a little unclear the cluster itself takes advantage of that capabiliy and if so how other than allowing an API for application code.

So, what's the difference between these products and which would I use? Could you please provide a link to technical documentation on the products?

Thanks in advance,

Mark

Answer 1

[Walter Roberson]

Parallel Server is aimed at high performance compute server environments, to run jobs on behalf of users, using multiple cores, possibly in a "batch" manner (but not necessarily.) The workers are not kept "live" between jobs, so start-up times for workers are comparable to starting up MATLAB itself.

Production Server is aimed at client/server interaction where it is important to reduce start-up times, but where the work involved per request is not necessarily all that much. Techniques are used to keep workers "live" for fast response time.