Global Red initially ran applications, up to 30 customer databases, and information for its business across the Asia-Pacific Region in two hosted data centers—one in Australia and the other in New Zealand. While the hosted infrastructure could support Global Red’s short-term compute requirements, forecasted traffic and customer growth coupled with the increasing availability of innovative cloud-based infrastructure services made moving to the cloud appealing to the ambitious firm. “By working with a global cloud provider that could deliver services such as the collection and processing of large data streams in real time, highly scalable compute resources, relational database services, and object storage, we could develop valuable marketing as a service IP,” says Kartik Rao, chief information officer at Global Red. “This IP would include real-time and predictive analytics, forecasting, and data collection and aggregation tools capable of capturing billions of events every month.”

Global Red also wanted to use the cloud to improve its agility and competitiveness by adopting continuous integration and deployment. One of Global Red's main business lines is a trading desk—Accumedia—that buys display advertising on behalf of its customers. For each advertising campaign, the business captures enormous volumes of data in a database and uses data-science algorithms to determine the most effective ad placements at any point in time. Adjustments are made in near real time to the buy based on what the algorithm shows.

“In our hosted data centers, it would take us up to six weeks to onboard a brand new client, create the databases, ensure we had sufficient infrastructure capacity for the project, and load-test to ensure we could scale as required. In a demanding, fast-moving market, this was too long,” Rao says.

“Now we can onboard clients with different data needs onto a single platform in a matter of weeks. This is great for our clients because we can focus our attention on extracting marketing outcomes and generating real ROI, rather than spend an enormous amount of effort building bespoke data ingestion pipelines,” he adds.