MQLess is a lightweight actor model framework on top of AWS Lambda.
Let’s look at one of the most basics issue with writing concurrent software. What do we do if two different functions need to update the same variable from different threads? If we just let them update it freely the state get corrupted. However, we all learned the solution. We synchronize access by using locks. The problem is, locks are slow and complex (deadlocks) and art only a few people know how to master.
These realities result in a no-win situation:
- Without sufficient locks, the state gets corrupted.
- With many locks in place, performance suffers and very quickly leads to deadlocks.
Still, you might say, that not my problem, my system is stateless.
Not quite right, you just moved the state to other services, which are stateful.
For example, how your software behaves if two (or more) operations try to update the same record in DB? Or two services try to update the same record in Redis? The typical solutions, optimistic & pessimistic concurrency, lead to the same problem as the example above, either to a corrupted state or slow software.
For some stateless applications, that is good enough, concurrency on same database row or logical entity is rare, and usually, they can live with optimistic concurrency.
For a high-performance low-latency application that is not enough, and a better solution is needed.
Enter the actor model.
The actor model was proposed decades ago by Carl Hewitt as a way to handle parallel processing in a high-performance network — an environment that was not available at the time. Today, hardware and infrastructure capabilities have caught up with and exceeded Hewitt’s vision. Consequently, organizations building distributed systems with demanding requirements encounter challenges that cannot fully be solved with a traditional object-oriented programming (OOP) model, but that can benefit from the actor model.
The actor model is similar to OOP, but instead of a class we have an actor, and instead of methods we have messages. Each actor has an address. If you know the address of an actor, you can send it a message. When you send an actor a message, it enters the actor mailbox. The actor is then processing the messages one by one.
MQLess implement the actor model on top of AWS Lambda.
Actor model frameworks like Akka and Microsoft Orleans use a cluster of servers to host the actors.
MQLess is serverless and clusterless, MQLess exploit Lambdas to host the actors. Each actor type is a Lambda function, handling the messages sent to the actor.
MQLess manage a mailbox for each actor address, and invoke the Lambda to process messages in the mailbox.