How it works

Alerts Processing

HeyNow operates on streams. If you are already familiar with reactive programming, getting to know HeyNow engine should be extremely easy. If not, please read this awesome The introduction to Reactive Programming you’ve been missing.

Lets assume that our example application, call it DeathStar, every now and then sends to HeyNow Event with core temperature information.

{
   eventType: "core-temperature",
   temperature: 2
}

We can describe this event stream on diagram:

coreTemperature: ---2----1--3----4--5---2-->

We want to receive alerts, when core temperature exceeds 3 enabling our engineers to start hyper-cooling systems preventing the whole construction from exploding.

First, we need to filter our stream:

coreTemperature: ---2----1--3----4--5---2-->
                 vvvvvvvv filter(>3) vvvvvvv
     dangerZone: ----------------4--5------>

Also, we don’t want to spam our engineers - the cooling systems needs couple minutes to kick in. So, after sending an alert we want to keep tool quiet for 15 minutes.

coreTemperature: ---2----1--3----4--5---2-->
                 vvvvvvvv filter(>3) vvvvvvv
     dangerZone: ----------------4--5------>
                 vvv throttleFirst(15m) vvvv
     alerts:     ----------------4--------->

All we need to do now is to subscribe the alerts stream and send email when stream emits.

HeyNow allows to express this flow in a simpler manner than in the examples above. When setting up new pipeline one can use diagram drawing tool to simply drag and drop appropriate operators and connect them using arrows.

Reactive Architecture

HeyNow is build on Reactive Architecture. This means that in addition to use Micro-Services Architecture it follows these four principles:

• Event-Driven
• Scalable
• Resilient
• Responsive

This is achieved by building whole solution around the queue system. Each Source, Operator and Sink is a different service, eg. we have MapOperator or EmailSink. Those services exchange Events, according to the Streams defined by users using queues. Each service can scale independently adapting system to the load. Because the stream processing is realised in near real-time by exchanging messages, system is responsive and cost-effective.