I read about stuff like CAP theorem, but what has me confused is when people say "with RDBMS, you have to look at every column in each row, which is less performant". Like what if in RDBMS I was doing a SELECT statement on just 1 column? Then couldn't it just go down that one column similar to Cassandra would?
And with use cases like Instagram, I imagine data like photo feeds and messages as relational, but they use Cassandra. I'm wondering why it's recommended for things like messaging and chat.
Basically I'm struggling to grasp wide column stores because it seems like RDBMS gives you a lot more when it comes to data modeling.
When I look at diagrams I either see some row key with columns that could just be an RDBMS row with primary key, or some JSON-like key values, so I'm having trouble visualizing the benefits.
Thanks!
Top comments (0)