Hmm... I wonder to what extent the bi-directional linking architecture of creates a computationally intractable problem... you've got to have a really rapid reactive process maintaining consistency... even before you run into cap theorem stuff in larger dbs
Conversation
Yeah it gets expensive fast. At human scale data sets though it's generally fine. Dunno how far sufficiently clever algorithms can take this though
1
1
I imagine the right way to approach such problems is to hold the largest subset of the graph possible in-memory on a single instance?
1
2
Yeah exactly. This article is nice
1
2
well... until some of the many rackscale computing projects get good enough
1
1
Mostly a problem for multiplayer, mobile, and getting quick load times.
Tractable but difficult problems
2
1



