Software engineering techniques for strong eventual consistent replicated data types

Distributed systems are shifting from centralized to decentralized architectures. In decentralized systems data is commonly replicated (copied) over different parts of the system. Replicating data in a system can improve efficiency and availability but complicates software development because concurrent updates can lead to conflicts.

Strong eventual consistency (SEC) is a consistency model where conflicts are impossible, availability is guaranteed, and consistency is eventually ensured. While replicated data types using SEC are interesting as they guarantee high availability and low latencies, most research so far has focused on their formal specifications. Programming support for SEC is lacking, which raises a barrier for using them in general-purpose applications. In fact, libraries for these data types generally do not handle the communication, memory management and security aspects. All of these aspects need to be manually handled by the application developer.

We propose a middleware that forms a foundation on which SEC
data types can be implemented.
It will provide the necessary run-time support for language mechanisms like replication, memory management, and security. By improving the implementation of SEC data types, application developers using them will be able to focus on their application logic rather than solving low-level replication issues.