Once you have created relationships within the dataset or register file, you can create cubes. When you define a cube, you specify the dimensions and measures and the types of pre-calculated aggregations it will contain.
Once you've completed the basic design, you can refine the cube by defining the partition, aggregation, and auto cache strategies, and setting advanced properties. Then you can schedule the build, get recommendations, analyze queries, review build summaries or data profile results, manage build instances, or add other jobs such as data profile or cache job.
All of the sources used to build a cube must be from a single connection.
As your business generates new data, you can add data to an already built cube with an incremental build. You can also create a sliding window and specify a range of data that you want to build into the cube.
You can delete cubes at any time, however, if there are reports built using the cube, they will no longer function, and you will receive a warning before you proceed.
There are two ways to view and work with a cube:
- logical view (default view)
- physical view
You can maintain the logical view of a cube to support business requirements and tune the physical view to support optimization requirements. The logical view shows all calculated measures and members.
Kyvos supports building cubes on both on-premise (Hadoop) and Cloud (Amazon S3 or Azure) platforms. You can also build cubes on a Snowflake or BigQuery source.
Before you get started
Before you create a cube, you may need to know about the schema, which describes how the tables are arranged in the database. Star, snowflake, multi fact, and single file schemas are common types you will encounter. They are different arrangements of facts (business data) and dimensions (the records in the fact table - for example, information about specific products or customers). A cube aggregates facts from various levels of a dimension provided in a schema. For example, you may have data about products, units sold, and sales amounts and match them up by date (month, quarter), by store, salesperson, or other combinations.
You can hover the cursor over specific dimensions or measures to see additional data about that item such as field name, node name, and function used. For count measures, count type such as accurate or approximate is also shown.
Kyvos allows you to create a data profile that provides guidance for cube design. Additionally, Kyvos provides cube recommendations to optimize the cube based on historical query data and the data profile. You can calculate aggregates when do you do the cube build to reduce query times.
You can also analyze queries created on the cube. The results are displayed in a dashboard in a separate tab in the browser. You can view query performance data and lists of mostly used dimensions and measures to aid you in optimizing your cube design.
- Defining a cube
- Creating expressions using MDX Linear syntax
- Creating cubes for GIS maps
- Formatting measures
- Defining drill down order in cubes
- Validating cubes
- Data security for cubes
- Configuring SQL schema
- Using cube properties
- Analyzing queries
- Cube build types
- Partition strategy
- Using aggregates
- Cube recommendations
- Changing underlying data
- Slowly changing dimensions
- Creating a data profile
- Scheduling cube builds
- Build history
- Job summary
- Exporting data lineage