3 ms·
Why not? I can think of one reason: SQL is generally poor at modeling tree structures. This may reflect itself in the application with comments without hierarch
by gregwebs 7y ago
Why not? I can think of one reason: SQL is generally poor at modeling tree structures. This may reflect itself in the application with comments without hierarchies (which is the case for StackOverflow and Quora), but that is a more minor part of the application. Similarly if you wanted to have a graph of relations between knowledge, graph databases might be better. But QA sites tend to just add tags to the questions, which is easy enough to model.
- B4TMAN 7y agoThe problem is most the interviews ask a question along the lines what if there are 10s of millions of rows. Would a SQL database be still preferable?
- kthejoker2 7y agoThe tradeoff is sharding and caching graph DBs is difficult in terms of the performance hit because it's harder to isolate queries to a single partition/subgraph - in some ways intersectionality is the whole point of graph vs relational. The overhead of a large scale distributed graph DB with redundancy and caching is significantly higher than a relational DB acting as a "poor man's graph DB"; and materializing your more expensive views and queries is, in the base case, a sufficient performance equalizer.