AWS Case Study: Actual Analytics

The company decided early that, due to the large and variable processing requirements associated with video processing, they were going to use a cloud platform to deliver their application. James Heward, Actual Analytics CEO, says, “It was important for us to have a solution that was mature, cost-effective, and global. Amazon Web Services [AWS] was the only offering that met these criteria. That consideration, combined with the great range of development tools, and their community, quickly made AWS the only choice for us.”

Currently, the company’s products are deployed as both a Software as a Service (SaaS) offering and an onsite product. Heward explains, “We use AWS to power all of the scalability, storage, and processing behind our SaaS offering. The entire application is deployed on AWS.”

The Actual Analytics team used four languages to build their solution, as follows:

• The low level video processing building blocks are implemented in C.
• The information extraction algorithms that use the building blocks are implemented in Python.
• The core of the application, including authentication, content management, and job scheduling are written in Java providing a REST / AMF API.
• The client is written in Adobe Flex.

Heward says that developing a new product specifically designed to run on AWS went smoothly. He adds, “The most valuable lesson we learned was to carefully understand the limitations of the AWS services we intended to use, because there were subtleties that could have impacted the design of the architecture if not thoroughly understood early on.”

The company may incorporate Amazon CloudFront into their products if content delivery latency becomes an issue; and they may use Amazon Elastic MapReduce to scale their back end processing.

Heward describes the results of using AWS: “AWS enabled us to bring our SaaS offering to market at a much lower cost than a traditional deployment, because our costs scale with user adoption, turning fixed overheads into cost of goods sold. Use of a traditional provider would have incurred an upfront cost of around £25k. Using AWS, we reduced the upfront costs to around £2k (testing usage). The upfront costs of delivering a platform with the scalability, security, and robustness of AWS would have killed us. Without AWS we would not have been able to bring the power of video content analysis to the world.”

To learn more, visit http://www.actualanalytics.com/ .