Im doing a project with php and mongodb. There is a requirement where the user needs to be shown a feed as in any social network system.
To make this happen im keeping a mongo document to each object (users,pages,etc) which has a unique id, each document is called a node similar to a node in a graph db. All the nodes are in the "nodes" collection in my mongo db. The relationship are stored with connection flow(direction) and referring id's (to nodes) in a collection called "rels".
When making queries i would simply aggregate on the "rels" collections and load respective nodes.
My question is how would this kind of a model affect when it comes to scaling ? because there will be uncountable amount of documents made on the "rels" collection. probably 10000's . Im pretty new to scaling in mongo. Is my approach wrong ? thanks.
To drive a nail through a wall you dont need a sledgehammer. MongoDB is not suitable for these kind of scenarios. Your use case where-in you have nodes and relationships between them can be best justified by using a graph database like Neo4J. The scaling aspect is yet another reason to choose Neo4J over MongoDB for as the connections/relationships can get complex with time as the number of nodes grow.