Company: Atos
Application: London Olympics’ spectator travel app with extreme and unpredictable loads
Server configuration: Elastically added servers with load
Reason for deployment: Needed easy-to-use data management platform that could handle unpredictable loads
Results: Software & development costs minimized, time-to-deployment minimized, handled all loads
Background
As the Worldwide Information Technology Partner for the Olympic and Paralympic Games, Atos were tasked with developing the Spectator Journey Planner, a public transportation application for travel planning between events. Predicting an unprecedented traffic load for the London games and working under a tight development schedule, Atos turned to ScaleOut Software’s in-memory data grid (IMDG) technology to implement a high performance and flexible infrastructure that could be deployed easily.
Challenges
By January 2011, Atos had reviewed the impact of the Beijing games and knew up to one billion visitors could visit the Olympic Games website, so any solution needed to be rapidly scalable. The demand that Olympic visitors could place on the Spectator Journey Planner required Atos to be able to bring online additional server capacity within 24 hours. As a result, Atos determined it would leverage the cloud to enable rapid deployment of new resources into the service. Use of SQL servers would have locked Atos into a specific, clustered, hardware infrastructure that was unattractive both operationally and financially.
Proof of Concept
Atos also considered other in-memory databases such as memcache, App_Fabric and Oracle TimesTen, but ScaleOut Software’s IMDGs and accompanying service offering were a perfect fit. Technically the grid’s object-oriented techniques were familiar to the Atos development team whilst ScaleOut Software’s product service was comprehensive and more attractive from a business perspective, in comparison, to other solutions for state management.
“During load testing, the performance of ScaleOut Software was excellent, and we did not encounter any issues even under peak application loads.”
–Steve Craddock, Technical Architect
ScaleOut Software demonstrated its ease of use as soon as the project began. The Spectator Journey Planner was based on another Atos service, Transport Direct, yet by implementing ScaleOut Software, existing application logic was migrated into the new deployment seamlessly.
The Spectator Journey Planner was adapted from an existing public transportation planner Atos had designed previously, Transport Direct, which was comprised of a stateful website built with Microsoft .NET and SQL Server. The design of Transport Direct, while being performant, did not allow for the rapid scalability required for the Olympics. Using a SQL Server for the state storage and session management would have locked Atos into specific hardware requirements, making it difficult to add capacity as demand increased.
“We were rather impressed at how easy the migration was to complete,” said Steve Craddock Technical Architect, Atos. “ScaleOut Software enabled an evolution of the [Transport Direct] service, rather than a revolution.”
Atos began to review possible solutions in December 2010. ScaleOut Software was very helpful in addressing its questions, providing quality support and configuration advice with a swift response. Atos began working on the Spectator Journey Planner in January 2011, and it was completed just three months later in March 2011 for review.
Results
The greatest advantage Atos gained from ScaleOut Software was the ability to easily scale out a high performance infrastructure as its load grew. By deploying ScaleOut Software into a cloud-based infrastructure, Atos could easily support additional servers with IMDGs as more Olympic visitors began to use the service.
“The performance, throughput and speed of ScaleOut Software was markedly better than our existing SQL platforms” said Steve Craddock. “During load testing, the performance of ScaleOut Software was excellent, and we did not encounter any issues even under peak application loads.”
The uncomplicated conversion process to ScaleOut Software enabled Atos to commit to an aggressive development schedule. ScaleOut Software employs a wide array of self-management features, making installation quick and simple. Also ScaleOut Software APIs use object-oriented techniques familiar to Java, .NET and C/C++ application developers to deliver a straightforward development cycle.
“The fact that ScaleOut made everything so easy was a big confidence boost early on in the project,” said Steve Craddock. “And the software included everything we could think of straight out of the box.”
ScaleOut Software was deployed on a small server farm of dedicated data grid servers running in conjunction with the application web servers. This configuration was chosen to allow administration of the web servers to be independent of the IMDG servers. The web servers running the Spectator Journey Planner were connected via a LAN to the IMDG servers using ScaleOut Remote Client. High availability was assured by ScaleOut’s replication and failure detection/recovery features.
Conclusions
Atos is currently moving forward with ScaleOut Software building on what it has learned and the confidence gained during the Olympics. To fully leverage the development work, Atos developed Spectator Journey Planner with a level of reuse in mind. Today, Atos is planning to push the technology from Spectator Journey Planner back into Transport Direct and is also assessing how to best to augment their web-based rail ticketing solution using the unique experiences gained from the Olympics, in combination with ScaleOut Software.
ScaleOut Software’s advanced in-memory data grid products help organizations meet a wide range of technical and business challenges. Whether you need a scalable storage solution for storing session-state or shopping carts on your e-commerce web farm or you have to perform lightning fast data analysis of financial data using your compute grid, ScaleOut Software’s breakthrough technology will eliminate bottlenecks and help you achieve your performance goals.