Investigation and Comparison of Distributed NoSQL Database Systems

HBase supports the BigTable data model [1] that was originally proposed by Google. Figure 1 illustrates this data model. Data are stored in tables; each table contains multiple rows, and a fixed number of column families. For each row, there can be a various number of qualifiers (columns) within each column family, and at the intersections of rows and qualifiers are table cells. Cell contents are uninterpreted byte arrays. Cell values are versioned using timestamps, and a table can be configured to maintain a certain number of versions. Rows are sorted by row keys, which are also implemented as byte arrays. Within each column family, columns are sorted by column names. Cell values under a column are further sorted by timestamps.