Creating Large-Scale, Highly Available OLAP Sites: A Step-by-Step Guide

Note: This paper was written for Analysis Services 2000, but most of the papers ideas still applies to versions 2005, 2008 and 2008R2.

Abstract This paper introduces various techniques needed to create large-scale, highly available online analytical processing (OLAP) sites using Microsoft® SQL ServerT 2000 Analysis Services and Network Load Balancing in a Microsoft Windows® 2000 Advanced Server infrastructure. High availability refers to the ability of a multiple-server site to withstand hardware or software outages that may affect individual servers within a site. A large-scale site is a site that has hundreds of gigabytes (GB) of data, making normal file copy operations ineffective. This paper provides step-by-step instructions that demonstrate how to build the servers and the supporting infrastructure for a reference site based on both thin-client and thick-client architectures. The architectures described are designed to: Protect the data of a multiple-server OLAP site and keep the site up and running during unplanned outages. Coordinate the allocation of processing resources during planned outages, such as data refreshes and complex extraction, transformation, and loading (ETL) cycles. Solve many of the staging-to-production operational problems that can occur in large OLAP sites.