Infectious Media Shaves Time Off Data Analysis and Server Maintenance with Google Cloud Platform
Founded in 2008 at the start of programmatic advertising, Infectious Media now delivers real-time online advertising campaigns for customers like Netflix, Sky and John Lewis. Based in London, Infectious Media delivers live campaigns in over 50 countries on any given day. The team is comprised of analysts, data scientists, engineers, digital media practitioners and business operations professionals.
Each month, Infectious Media processes more than 40 billion ad decisions and 100 gigabytes of data across 60 different sources. The team has always cared about making sure data storage and compute systems aren’t affected by spikes in data requests. With its previous cloud platform, Infectious Media experienced stalls for several minutes, or even worse — crashes — during times of heavy demand. Infectious Media then moved to a non-cloud managed server environment, which solved most of the data performance problems, but the cost continued to increase as the company scaled. They realized they needed a more sophisticated cloud system that could manage the demands of a high-frequency trading system and could scale in a cost-effective manner.
Infectious Media chose Google Cloud Platform for its ability to scale the hosting and compute capacity without overburdening developers with maintenance tasks. The team migrated from its managed services system to Cloud Platform in just one week with no downtime. “Cloud Platform offers a stack of services which allows us to coordinate our efforts and keep our data in one place,” says Daniel Bowtell, head of TechOps at Infectious Media.
Google Compute Engine runs the main billing platform and manages requests for advertising by clients, while Google BigQuery provides customer data analysis and reporting, including insights into which ad impressions convert into customer leads. Google Cloud SQL also manages the business rules database.
With Cloud Platform, Infectious Media is more efficient and no longer worries about capacity and latency issues. The company handles more data requests per second at the same price of its old server environment. Before using Compute Engine, Infectious Media processed 35,000 data requests per second, and today they’re processing four times that amount of hosted data at the same price.
The company is able to host more data and provide more complex insights into what the data means. When the company used a columnar SQL database, processing customer data for each project took 20 minutes. Now with BigQuery, it takes 20 seconds, freeing employees to dig into the data with more complex calculations. “Everything happens faster with Cloud Platform, and when it comes to data analysis, the difference is impressive,” Bowtell says.
Since the platform is simple to use, more employees can use the data tools, rather than limiting access to a few data technicians. This leads to more detailed reports in BigQuery that take less time to produce. Infectious Media now can also scale services up or down much faster. It used to take each machine an hour to shift from using 20 machines at peak times, to six machines to accommodate the slower overnight load. Today the entire process takes about two minutes using Autoscale in Compute Engine.
This combination of Compute Engine, BigQuery and Cloud SQL helps Infectious Media scale efficiently while processing more data faster, ultimately helping the team enjoy cost-effective capacity without ever having to sacrifice speed.