Structure your data using BigTable/HBase

How I Learned to Stop Worrying and Love Using a Lot of Disk Space to Scale:

BigTable or Hbase is essentially a distributed hash table. To structure your data using such databases, you have to adapt your mind against certain principles you've used a long time in RDBMS.
1. Denomalization instead normalization. Using duplicates to maximize concurrent reads.
2. Compute attributes at write time to prevent applications from iterating over large bulk of data.
3. Create large entities with optional fields to avoid any joins.
4. scale for reads and taking the pain on writes.