Transcription
In existence in the past, we had something called the relational database. But the problem with the relational databases is you had to define your columns. You had to do it ahead of time.
In a relational database, you have to go to the database engine and you have to kind of do a migration. That usually happens at compile time, not at runtime.
So then enters document databases such as database. Document one has like three or three attributes. It's first name, last name, email. Document two has only one. Any moment I can add, at any moment I can remove. I don't have to have a fixed, what I call, a fixed schema for my data.
Living in a new world order, the AI world, a lot of things are going to go now dynamic. And therefore, the best bet is something like a document.
>> [music] >>
But you want to also store arbitrary, not only the like documents, but relationships between documents, what we call graph databases. Neo4j is a good example of that. Also want to store edges, aka relationships, and I should be able to query them very, very fast.
There is a special type of data, which is we call time series data. World we currently live in. There is no database that can tackle all of these use cases at once and play nicely with AI.
But where does this variable data live? That's the gazillion-dollar question.