Raw Binary Tree Database or MongoDb/MySQL/Etc ?

Question

I will be storing terabytes of information, before indexes, and after compression methods.

Should I code up a Binary Tree Database by hand using sort files etc, or use something like MongoDB or even something like MySQL?

I am worried about (space) cost per record with things like MySQL and other DB's that are around. I also know that some databases even allow for compression, but they convert to read only tables. These tables/records need to be accessed and overwritten with new data fairly often. I think if I were to code something in C++ I'd be able to keep the cost of space per record to a minimum.

What should I do?

Answer 1

There are new non-relational databases that are becoming popular these days, that specialize in managing large-scale data.

Check out Hadoop or Cassandra, both of these are at the Apache Project.