BI Publisher Architecture & Data Model Basics

BIP's architecture cleanly separates data from presentation, which is the single most important architectural idea to articulate. A Data Model defines where the data comes from (SQL query, an OBIEE subject area via logical SQL, a web service, an Excel/CSV file, or even an existing OBIEE analysis) and shapes it into an XML data structure — this XML is the contract between data and layout. A Layout Template (built in RTF via Microsoft Word using the BI Publisher plugin, in Excel, in native BIP Layout Editor, or as an XSL-FO/XSLT template for advanced cases) defines exactly how that XML should be rendered — fonts, tables, page breaks, headers/footers, barcodes, and precise positioning of every element.

At runtime, the BI Publisher engine merges a chosen data model's XML output against a chosen layout template to produce the final output in whatever format is requested (PDF, Excel, Word, HTML, PowerPoint, or plain text/EDI for system-to-system feeds). Because data and layout are independent artifacts, the same data model can drive multiple different layout templates (an English invoice template and a French invoice template, both fed by the identical underlying XML), and the same layout style can be reused across different data models if the XML structure matches.

A sharp interview answer draws the line clearly: OBIEE Answers is for ad-hoc, interactive, web-rendered analysis; BI Publisher is for fixed-format, often high-volume, print/document-quality output — and many real deployments use both, side by side, for different audiences and use cases.