Do we have to always ensure there is enough memory to contain the dataset?

Question

Given that Memgraph is in-memory first, does this mean that we must always ensure there is enough memory to contain the entire dataset, and if it consumes too much memory, it will die rather than spill over to disk?

Answer 1

A user has to ensure that there is enough memory for the entire graph, but Memgraph won't die if too much memory is consumed because it has a memory tracker that stops the execution. For example, if there is no more available memory and you try to create a new node, Memgraph will create a snapshot and save it to the disk, but you will need to increase the amount of memory in order to load the whole graph.