This is a WebFOCUS-integrated database based upon technology from InfoBright (previously known as Brighthouse). Information Builders calls Hyperstage a "load-and-go," high-performance, highly scalable database that is simple to implement, use, and maintain.
Like other database adapters, Hyperstage is an add-on product integrated into the WebFOCUS Reporting Server, uses standard metadata, and is administered through the server console.
While relational databases store data organized into two-dimensional tables containing rows and columns, Hyperstage is designed to store together serialized values of each particular column. Rather than retrieve entire rows, a columnar database only reads the necessary columns.
This approach has proven useful for BI systems which need quick access to small subsets of data. Columnar databases in generate are optimized for analytic queries which ask questions about the data (e.g., aggregations and trends) rather than displaying records from the data.
Hyperstage organizes columns into "Data Packs" which are compressed to reduce the physical size of the database. While columnar data compressions of 10:1 ratios are common, an IB resource says one client obtained a 45:1 compression rate.
For faster retrieval, Hyperstage caches the Data Packs into memory. Hyperstage optimizes data access using an in-memory metadata layer known as the "Database Knowledge Grid." This feature helps eliminate the need for typical performance improvement techniques such as indexes, partitions, and materialized views.
Hyperstage uses an open-source relational database (either MySQL or PostgreSQL) and adds its own layers for improved data storage and query optimization. The WebFOCUS Hyperstage adapter will transform BI requests into optimized SQL queries. Queries that cannot be optimized by Hyperstage are handled by the relational database engine.
While the WebFOCUS language does not currently support a "HOLD FORMAT HYPERSTAGE" option, we should be able to easily replicate this feature with a small custom utility.
For example, many of my clients have nightly batch processing to load special BI databases for fast dashboard access. For Hyperstage, we might simply take some of those existing flat import files and use them as the input to a Hyperstage bulk load process.
Using the WebFOCUS language, we can develop a simple procedure to read the flat file's metadata and generate two SQL routines:
• CREATE command to initialize a Hyperstage structure (using ENGINE=BRIGHTHOUSE)
• LOAD DATA INFILE to bulk-load the Hyperstage structure
You can download my prototype of this WebFOCUS utility from my GitHub account.
Do you see places where Hyperstage might be of use within your organization? Let me know if I can be of service.