Reactive Systems

Goal #

Provide an experience that is responsive under all conditions. Note that reactive programming is not the same as reactive systems.

This requires:

Ability to scale from 10 users to million of users.
Consume solely the resources required to support the current work-load.

Reactive Principles #

Systems that apply the following principles are considered reactive systems (see more here ).

Responsive #

Always respond in a timely manner.

Resiliency #

Isolate failures on single components - Similar to how a boat is designed.

Elastic #

Keep responsive specially when the system load changes which provides a more efficient usage of resources.
Reduce contentions.
Avoid central bottlenecks.
Predictive Auto-Scaling policies.

Message Driven #

Losse coupling
Isolatation - Kinda disagree with loose coupling, as the systems still will still depend on 3rd party behavior regardless of the communication medium.
Resources are only consumed while active

Avoce all, make sure that the systems is not actively waiting so that it can do something else in the mean time which leads to a Async and non-blocking messages.

Git as example #

Asyncronous and non-blocking as the work is submitted through PR and I can keep on working locally with no interruptions. Message based as it is basically “please review this”.
Resiliency as each user has basically a copy of the whole repository locally. The local machine is isolated from the remote.
Elastic has we can have multiple copies and does not cause problemas having that repository sync on that many machines.
Responsive.