KijiSchema’s data model is an extension of HBase’s columnar storage model. Like HBase, KijiSchema’s tables contain many rows; each row contains many columns. A given cell identified by a row and column can hold many different timestamped values, representing the changing values of the cell over time. Columns are grouped together into column families, which provide namespacing for columns that store related data. Families are grouped into locality groups, which help store related families physically close to one another.

What Kiji calls a "locality group", HBase calls a "family". The Kiji column "family" allows you to choose the logical grouping and namespace of columns separately from the physical configuration of how the data is stored. In Kiji, multiple families can belong to one locality group.

As in HBase, rows in Kiji tables can have an arbitrary number of columns. Individual rows may have hundreds or thousands (or more) columns. Different rows may not necessarily have the same set of columns.

Unlike HBase, each cell in a Kiji table has a schema associated with it. Schemas in KijiSchema are versioned. The schema and layout system is described in greater detail in Managing Data.

Entity-Centric Data Model

KijiSchema’s data model is entity-centric. Each row typically holds information about a single entity in your information scheme. As an example, a consumer e-commerce web site may have a row representing each user of their site. The entity-centric data model enables easier analysis of individual entities. For example, to recommend products to a user, information such as the user’s past purchases, previously viewed items, search queries, etc. all need to be brought together. The entity-centric model stores all of these attributes of the user in the same row, allowing for efficient access to relevant information.

The entity-centric data model stands in comparison to a more typical log-based approach to data collection. Many MapReduce systems import log files for analysis. Logs are action-centric; each action performed by a user (adding an item to a shopping cart, checking out, performing a search, viewing a product) generates a new log entry. Collecting all the data required for a per-user analysis thus requires a scan of many logs. The entity-centric model is a “pivoted” form of this same information. By pivoting the information as the data is loaded into KijiSchema, later analysis can be run more efficiently, either in a MapReduce job operating over all users, or in a more narrowly-targeted fashion if individual rows require further computation.