I know redis graph maintains the adjacency relationship matrix in memory and there are many articles mentioning that its suitable only if your graph fits in the memory.
I'm interested in creating a social graph containing users with different relationships as well as posts that each user can make on which others can comment and likes on. As with any new app, I'm anticipating to start with few users about 100k (fewer nodes & relationship to begin with) & grow over time.
While in theory its possible to implement in RedisGraph, I'm unable to judge or account for future growth of the graph and know whether it can still fit in memory & perform at the same level.
Is RedisGraph suitable for such usecases or is it more suited towards somewhat static graph?
RedisGraph uses sparse matrices to track graph structure in memory, it is the usage of sparse matrices that allows RedisGraph to contain large graphs.
It is common for natural constructed structures e.g. social network to be sparse i.e. the number of formed connection out of the total number of possible connections is extremely small, thus keeping track only after the "active" connections is cheap.