20 / 34

HBase Design Guidelines

Not able to play video? Try with youtube

Transcript: Now, let's revisit what we have learned so far about the data model of HBase:

  • Table is indexed based on the Row Key meaning the rows in the table are sorted or ordered based on the row key.
  • A row key can contain any data not only string or numbers. It could be a huge binary file such as a movie or a music file.
  • While adding a row, it is a must to provide the row key.
  • Similar to adding rows to the table, we can keep adding columns. But a column has to be added to a column family.
  • Also, the column has a column name which can contain any data of any size.
  • Atomicity of a transaction is guaranteed only at a row level. Not on the transactions involving multiple rows meaning, we can make multiple modifications to a row in a single transaction but we cannot make modifications in a single transaction involving multiple rows.
  • Column families have to be defined upfront at the time of table creation.
  • All data model operations return data in sorted order: row key, ColumnFamily, column qualifier, timestamp which is in descending order.

Loading comments...