Our client initially retained Amadeus Consulting CTO, Steve Loper one of the top .NET/SQL architects in the country, as a consulting database architect. Steve was soon elevated to a team leadership role, and his fulltime, onsite engagement lasted for 18 months. Steve applied his expert abilities and knowledge to the most critical software development projects and was selected to lead an internal team of 20 software developers on three continents. This entailed managing work groups across multiple languages, cultures and time zones. Steve traveled to Moscow and to Kiev, Ukraine during the turbulent presidential elections in 2005.
Among the projects led by Steve Loper was the design of a new "Metadata Database" to manage and continuously update all the detailed information about the exchanges, markets and investment instruments handled by the extensive trading platform. This information was maintained in multiple separate data repositories requiring considerable manual intervention for updates and hindering their ability to keep current this critical metadata.
Amadeus Consulting provided the expertise in SQL Server architecture and business process optimization to architect the Metadata Database. This entailed the thorough definition of the logical data requirements driven by the conceptual business processes. Next, the physical database design was architected to consolidate the existing disparate data repositories into a single SQL Server database. Since the company maintains data centers near major market exchanges in North America, South America, Europe and Asia, the design included a replication strategy to optimize performance and response time for customers around the world.
To provide the necessary user interface and functionality, Steve also architected the system's data maintenance application allowing internal staff to manage the collection and proper usage of data gathered from the market exchanges. Steve conducted extensive interviews with department heads across the organization to ensure that all data sources were well understood. The application was developed as a Windows Forms application written in C#/.NET
On the next key project, Steve Loper served as the technical team lead for the design and development of the company's core data distribution network or "ticker plant." This network manages the normalization and distribution of dozens of custom real-time data feeds at a combined rate of over 100,000 messages per second incoming from worldwide exchanges. This involved the aggregation and normalization of real-time data and archiving specific subsets to be made available as historical data.
Steve worked closely with senior management evaluating and modifying existing and new designs. The system design includes parsers which read, filter and normalize incoming data from various custom data feeds. The data is then streamed to the real-time servers, which filter the data sent to each individual client according to their needs. In addition, a Master Server farm was configured to listen to all the parsers for incoming data and aggregate and store historical data.
Server redundancy was another important requirement, and Steve designed and developed the parser architecture to include geographically distributed clusters for instantaneous, automatic fail-over. This was accomplished using Microsoft Cluster Services with a Custom Cluster Resource DLL.
Steve Loper was instrumental in ensuring the successful development of a new technology platform and provided invaluable management and technical capabilities to in achieve their business goals. The team led by Steve significantly exceeded the benchmark metrics for end-to-end system latency time and performance goals.
The Metadata Database was successfully deployed resulting in a 50% reduction of data maintenance costs and other improvements in operational efficiency. The initial implementation of the Real-Time Data stream was completed and deployed to a first set of customers.