Well, but in a column-oriented database you would store your time series data in columns. Cassandra is "row-oriented" and you store the time series in rows. In both cases the data ends up sequentially on the disk, which is what you want.