Scaling and high availability for OutSystems Platform servers

Front-end

In a cluster scenario, the Front-end role scales best horizontally. Balancing the load across nodes will provide failover and several levels of disaster recovery, depending on the implementation.

The design can range from a segmented distribution of Front-ends that contain small incidents, to a geographical distribution aimed at containing high-risk events.

To properly scale your Front-end horizontally, Outsystems recommends that you decide on a suitable set of hardware specs for the Front-end server, and replicate that configuration on subsequent Front-end machines.

Deployment Controller

This service can only grow vertically since it has no cluster capability. Adding CPU cores, memory and increasing disk I/O will provide faster compile times for large/very active teams.

In a Production environment, it is only in use when new application versions are published. In a Development environment, it is used every time a developer publishes an eSpace, application or solution.

Scheduler

This role can be activated in any node with the Front-end or Deployment Controller roles. The OutSystems Platform will distribute the load generated by timer executions across all nodes that have this role, within the same OutSystems zone.

With heavy analytics timers, the pressure on shared Front-end and Deployment Controller nodes can negatively influence their performance. In this case, a dedicated Scheduler node or nodes are recommended.

LifeTime

LifeTime scales horizontally like any Front-end, mostly for high availability reasons, since it simply consists of a very light management console.

You should set up a dedicated node with, at least, the recommended hardware requirements for a Front-end role.