Types of CRDT

CRDTs (Conflict-free Replicated Data Types) can be classified into two main categories based on how they handle concurrent updates and ensure eventual consistency:

1. Operation-based CRDTs

In operation-based CRDTs, each update to the data structure is represented as an operation. These operations are commutative and idempotent, meaning they can be applied in any order and multiple times without changing the result. Operation-based CRDTs use a combination of operations and metadata (like timestamps or unique identifiers) to ensure that concurrent updates from different replicas can be merged correctly without conflicts. Examples of operation-based CRDTs include:

• Counter: A counter CRDT allows increment and decrement operations to be applied concurrently across replicas, ensuring that the final count is consistent across all replicas.
• Set: A set CRDT supports add and remove operations for adding or removing elements from a set. Concurrent add and remove operations are merged to ensure that the set remains consistent across replicas.
• List: A list CRDT supports insert and delete operations for modifying a list of elements. Concurrent insertions and deletions are merged to maintain a consistent order of elements across replicas.

2. State-based CRDTs

In state-based CRDTs, each replica maintains its state independently and periodically exchanges its state with other replicas to ensure eventual convergence. State-based CRDTs use a merge function to combine the states of different replicas in a deterministic manner, ensuring that all replicas eventually converge to the same state. Examples of state-based CRDTs include:

• Grow-only Set: A grow-only set CRDT allows elements to be added to the set but not removed. Each replica maintains its set of elements, and when exchanging states with other replicas, it merges the sets by taking the union of all elements.
• Last-Write-Wins Register: In this CRDT, each replica maintains a register with a single value, and updates to the register are timestamped. When merging states, the update with the highest timestamp is chosen, ensuring that the most recent update wins.
• Observed-Remove Set: This CRDT extends the grow-only set to support removal of elements. Each removal operation is accompanied by a tombstone indicating that the element has been removed. When merging states, tombstones are used to remove elements that have been deleted from one replica but not from others.

In the Distributed system, ensuring data consistency across the different nodes is a very critical challenge to solving this complex problem here comes out concept of Conflict-free Replicated Data Types (CRDT). CRDT enables multiple replicas of data to be updated independently and concurrently without the need for complex synchronization protocols.