Thomson Reuters is using the Intel fasterLAB to improve the performance of its integrated Reuters Enterprise Platform which draws on both real-time and static reference data to create a golden copy.

The Reuters Market Data System (RMDS) has already been through the lab. Reuters Enterprise Integration Engine (REIE) is built on data transport technology from M35 which has also been performance tuned in the Intel fasterLAB. REIE integrates real-time market streams with non real-time data and provides a data model where all the information is delivered. Then the information is made accessible through a suite of easy-to-use API calls and Web services calls, explained Nigel Matthews, Global Head of Product Management, Enterprise Platform - Data Services at Thomson Reuters. The company also provides a number of value added data services like exception handling, data reconciliation and symbology.

“We haven’t put the Reuters Reference Data System through the Lab yet because we are still in discussions about how to do it. We are looking at a scenario that would test all three Reuters services together.”

The three components of REP have different performance requirements. For RMDS, banks need high throughput and low latency. The integration layer, REIE, is scored on how many present values a user can calculate and push through the system, while RDS is measured by database performance – how many updates can a user enter into the database or publish from the database.

“It is interesting from our perspective that there is no one performance criteria but a whole range we have to consider because we are pulling together very disparate types of technology,” said Matthews. “RMDS is realtime, has no storage and is stateless while Reuters Reference Data consists of large files processed in batch and focused on storage and persistence. So in our work with Intel, we are looking for a whole range of characteristics to measure and improve.” Clients want to see numbers rather than estimates, he added, and providing a benchmark from an independent organization like the Intel fasterLAB.

Like its competitors in reference data, Reuters collects data in its Reference Data System, undertakes quality checks, verification, normalization and checks tolerances to create a golden copy. But its solution goes beyond that to integrate real-time data streams with the golden copy, link in-house applications and provide the results to upstream processes, he added.

“Our story isn’t just data management but integration, orchestration. We don’t believe we have a competitors who offer our type of complete end-to-end solution which can automate downstream business processes.”

The Reuters solution will be useful for risk management, compliance, and portfolio management where users need both reference data and real-time data.

“We are talking to clients who say that while there might have been a time when you could separate real-time and non real-time data, these days with the demands of modern trading systems or risk systems, the lines are blurred. For example, many organizations want to perform VAR in real-time. The batch approach is becoming problematic because you don’t have enough hours in the night to perform the calculations, and the results aren’t available during the trading day when they have the most value.” Other major drivers are counterparty risk exposure and portfolio valuations which require integration of real-time and static data.

The three components of the Reuters Enterprise Platform have been deployed for years and the combination is picking up interest and some early customers, including one in Austria, he added.

(No Ratings Yet)  Loading ...

Filed under: Issue 6 - Autumn 08