Tejas Jadhav

The Actor model

April 22, 2016

The Model

Have you heard of the Actor model in context of software architecture before? If you have, then continue reading for a quick revision; else, be prepared to understand one of the most amazing software architectures ever thought of. The Actor model somewhat works like this:

  • There’s a guy called as the Actor who is extremely good at doing one specific thing. He also has some good letter writing skills.
  • Imaging this guy is kept alone in a closed room without windows and doors, but just a small letter slot.
  • All this guy has to do is, whenever he gets a letter from the letter slot, he reads it, writes his reply in a new letter and sends the new letter out from the same letter slot.
  • The guy doesn’t know from whom he received the letter and doesn’t even know who will read his letter. All he has to do is respond to the incoming letters. If the content of the letter is intended for him, he can reply. Else, he may just ignore.
  • The guy who is sending him the letter can be another Actor sitting in some other closed room like him or any third party agent.

This is it. This is called the Actor model. This model is pretty close to pure functional programming if you relate Actors to functions. In fact, one of the high functional programming language, Erlang, is completely based out of the Actor model.

What’s so great about this model

Imagine, individual entities doing their own job without having a dependency on any other entity and exchanging data through strict message passing. Doesn’t that sound great? Since Actors are restricted from communicating directly with each other, we can ensure that communication between them is transparent. Another interesting thing is, it’s more of a Publisher-Subscriber (PubSub) architecture where the Actors send the message out in the wild. Other Actors can read this message and if they find something in that message that is related to their job, they can reply back with another message and it all across.

As Actors perform like individual entities, they can be put anywhere as long as the messaging channel is available to them. Think of this in context with servers distributed across a huge network. Each server is an Actor and the network is the messaging channel.

I mentioned earlier that Actor model strongly resembles functional programming. Here’s a small piece of code to demonstrate that,

def apple_actor(message):
    if message.contains('apple'):
        return 'An Apple a day, keeps the doctor away.'

def length_actor(message):
    if len(message) > 10:
        return 'That was some really some long message.'

The apple_actor is concerned with only those messages which have the word apple in them. Similarly, length_actor is concerned with only those messages which are more than 10 characters. So, despite the fact that they may receive all the messages, they will respond to only those with whom they are related to. In return, they will respond with another message which will be broadcast everywhere.

For instance, length_actor can respond to the response of apple_actor since it exceeds the 10 character count. This is how Actors actually communicate without knowing the source or destination.

Since these entities act to certain messages, they are called Actors.

Written by Tejas Jadhav full time software engineer

© 2021 Tejas Jadhav