Many software developers appreciate Matlab for the ease with which algorithms can be developed but do not consider Matlab as a tool for the final deployment of their applications. However there are great benefits in using a single tool for the full "devops" cycle of experimentation, design, coding, testing and (scaled) deployment.
Typical requirements that apply during the deployment phase are:
- must run stable, unattended and 24/7;
- must integrate seamlessly in a service oriented architecture;
- must scale easy and royalty free;
- must run in the cloud or on private Linux servers;
- must integrate with orchestration tools like Kubernetes and Docker Swarm.
Over the last two decades Modelit has gained experience with Matlab based algorithms that run on a continuous basis. Examples are our sites for short term travel time prediction or accident analysis.
Year over year we have improved our software tools and procedures that allow us to meet the requirements above.
Our latest additions are:
- The development of the Matlab toolbox "embedded httpserver". This toolbox contains a server object that allows the user to specify a callback that will be triggered by incoming requests. The toolbox also supports SSL and authentication.
- We have documented procedures for deploying single- and multi container web-oriented applications using Docker and Docker compose.
We have tried our best to design the toolbox and document related procedures in such a way that Matlab developers can independently reproduce all required steps.
However it is neither be doable nor productive to exhaustively document all knowledge that was gained over the years. Nor is it possible to anticipate all applications that users might have in mind. For this reason we offer consulting services to help you set up your first Matlab based web project. Please contact us for further information or a demo.