Solving the Distributed Snapshot Problem

Chandy-Lamport algorithm

The Chandy Lamport algorithm solves the consistent snapshot problem in a distributed system.

Idea

The algorithm is based on the following main idea: a marker message is sent between nodes using the available communication channels that represent an instruction to a node to record a snapshot of the current state.

Working

The algorithm works as follows:

• The node that initiates the protocol records its state and then sends a marker message to all the outbound channels.

Importantly, the marker is sent after the node records its state and before any further messages are sent to the channels. ...

• When a node receives a marker message, its behaviour depends on whether the node has already recorded its state (while emitting the mark previously)