Polyglot persistece with a graph database for relationships is a good ideia? -

i know if worth idea of use graph databases work relationships.

i pretend use relational database storing entities "user", "page", "comment", "post" etc.

but in cases of typical social graph based workload, have deep traversals relational not deal , involves slow joins.

example: comment -(made_in)-> post -(made_in)-> page etc...

i'm thinking make this:

example:

user id: 1

query: get followers of user_id 1

• query neo4j outcoming edges named "follows" node user id 1

• with list of ids query them on users table:

  select * users user_id in (ids)

is slow?

i have seen question is idea use mysql , neo4j together?, still cannot understand why correct answer says that not idea.

thanks

using neo4j great choice of technologies application yours, requires deep traversals. reason it's choice two-fold: 1 cypher language makes such queries easy. second deep traversals happen quickly, because of way data structured in database.

in order reap both of these benefits, want have both relationships , people (as nodes) in graph. you'll able friend-of-friends query follows:

start john=node:node_auto_index(name = 'john') match john-[:friend]->()-[:friend]->fof return john, fof

and friend-of-friend-of-friend query follows:

start john=node:node_auto_index(name = 'john') match john-[:friend]->()-[:friend]->()->[:friend]->fofof return john, fofof

...and on. (same idea posts , comments, replace name.)

using neo4j alongside mysql fine, wouldn't in particular way, because code more complex, , you'll lose time hopping between neo4j , mysql.

best of luck!

philip