Yes, it’s true. First time I had this in my mind, numbers were clearly against me. It was the time of the original Raspberry PI, but now the 3 is out since a bit, and I reconsidered the original question…

Let’s clarify, immediately. We’re not talking of a portal used by hundred of users eagerly and simultaneously, but of some modern web application, used internally to my office, with a really low probability of simultaneous overloading by more than 4 users.

With “modern” I mean a webapplication served by Tomcat 8 on Java 8, built on the “classical” stack of Hibernate / Spring /Spring MVC, with Single Page Client based on Angular, and MySQL as persistance layer.

Previously, those applications were hosted in my virtualization server, that is getting old and unused. The virtual machines will go used only on request, while those webapps are supposed to be always online. Keeping switched on a “tower” with noisy fans (that are getting always more noisy), power draining hard disks, was no longer attracting to me.

So I gave a try to the RPI3+.

Well, operation SUCCEEDED. And I have the sensation that performances are even more interesting of the former architecture.

In the operation I took care of

  • The system setup on the RPI, carefully monitoring temperatures of the device
  • Setup of the DB, the App server and apps themselves (including an asyncronous scheduled mailing component)
  • Setting up of course all is related to backup and monitoring of the device

Stay tuned, I will write down in these articles the relevant part and discoveries made during the migration!!

Using a Raspberry PI to serve Java/Spring/Hibernate web applications