Gordon Food Service Improves BI with Data Replication

Executive Background

Gordon Food Service Canada (GFS), a premier foodservice distributor, turns to Syniti for its data replication needs in the implementation of an enterprise-wide business intelligence (BI) solution for all divisions of GFS Canada. The purpose of the GFS Canada BI project is to implement the necessary data warehouse processes, tools and techniques to allow product item, vendor and customer margin and profitability analysis for the sales and marketing business areas.

About the Company

Gordon Food Service (GFS) Canada is a premier food service distributor. With over 100 years of business behind them, in 1994 GFS entered the Canadian marketplace through the merger of some of the more significant businesses in the foodservice industry. Their explosive growth requires GFS Canada to handle more products, provide more services and satisfy more customers’ needs than ever before. GFS Canada turned to Syniti for their data replication needs as a key component in the delivery of an enterprise-wide business intelligence solution for all divisions across GFS Canada.

Project Background

Stuart Bestbier, Project Manager for the Delivery of BI Solutions at GFS Canada, manages the design, development and implementation of an enterprise-wide BI solution for all divisions of GFS. The project includes selecting and implementing the necessary BI software and supporting infrastructure. The BI solution will support analysis, reporting and decision-making across all divisions. The BI project needs to manage approximately 50 tables with between 20 and 230 columns per table, and between 150 and 26 million records per table.

The BI project needs to manage approximately 50 tables with between 20 and 230 columns per table, and between 150 and 26 million
records per table. The overall number of new transactions per month runs at around 3 million, translating to roughly 100,000 primary
business transactions per day with many hundreds of thousands of supporting transactions every month.

Selection Criteria

The primary criteria used to select the data replication tool were:

• Minimize the impact to the mission critical production source system and have no negative impact on end user access to production data
• Have flexibility to automatically stop and restart the replication process in order to force a “point of quiescence” in the Data Warehouse extract process • Capture only the changes to source data (i.e. the deltas) from one Data Warehouse extract to the next
• Have a 100% reliable mechanism for capturing changes to source data, including a verbose record of all inserts, updates and deletes
• Product performance: GFS Canada data volumes require a fast, reliable data replication capability

Product Functionality

Bestbier’s project uses continuous one-way mirroring of source data to the Data Warehouse Staging Area with controlled stops and restarts. The use of Syniti allows them to overcome the following challanges:

• Removes the need to write custom code to identify source data changes from one day to the next
• Removes the need to read source data tables directly thus minimizing potential impact to end users

Results

• It enabled them to minimize impact to their mission critical production source system, while generating no negative impact to end users attempting to access production data
• Syniti is flexible enough to allow the BI project to automatically stop and restart the replication process as needed
• Syniti replication setup allows project developers to capture only the changes to source data from one Data Warehouse extract to the next. The mirroring mechanism captures all changes to source data, including a verbose record of all inserts, updates and deletes
• Syniti met the project requirements for a fast, reliable data replication capability for large volumes of data.

Vendor Results

Bestbier reports that technical support has been very good so far (mostly during the pre-sales cycle when the team was coming up to speed with the product features and capabilities). In addition, documentation met their needs, although more detailed examples to help out with custom scripting would be useful.