18 March 2018

STP - better by design

18 March 2002  |  1244 views  |  0 Philip Czachorowski - CSC

As part of its IBM alliance programme, CSC built a GSTP concentrator prototype in order to demonstrate how straight-through processing (STP) solutions can be built with message queuing and message broker technology using MQSeries® and MQSeries® Integrator (MQSI). Philip Czachorowski reports the findings.

The industry initiative to implement STP is being driven by the need to reduce cost as much as by the move to T+1. Financial services institutions (FSIs) are facing four technology inhibitors as they consider STP: connectivity, multiple technology integration, high-latency batch processing, and the need for data aggregation and distribution.
Global straight through processing (GSTP) provides an opportunity to fundamentally change the operating models for each industry participant. Asset managers can eliminate today's dependencies and achieve significant cost savings. Broker/dealers can drive down infrastructure costs and ensure their survival. Custodians can protect their franchise by achieving competitive advantage through advanced technology and global service offerings.
CSC built a prototype of a GSTP concentrator to demonstrate how STP solutions can be built with message queuing and message broker technology using IBM's MQSeries® and MQSeries® Integrator (MQSI). A realistic simulation of a GSTP workload was implemented and tested under high message rates to gain an in-depth understanding of the STP design and performance issues.
Our key findings are:
• Message queuing and brokering products, like MQSeries and MQSI, make it technically feasible to integrate a wide variety of STP applications, both internal and external. These products provide the core messaging and integration services that are key to implementing STP solutions.
• Our results show that, with the appropriate design, MQSeries and MQSI can support the largest FSIs and their trading volumes. An impressive throughput rate of over 300 GSTP messages per second with persistence was attained on an IBM RS/6000 S80 using a single MQSI broker. This volume corresponds to about ten trades per second. Based on the test findings, cost effective design strategies were developed for horizontally scaling solutions to virtually any message volume.
• Simplifying the implementation of STP solutions is critical toward meeting the business goals of reducing costs, improving business flexibility, and increasing responsiveness to change. The prototype demonstrated how to architect and design a STP solution that meets these goals, while still building a robust system. For example, a dynamic message routing technique was implemented, using XML, which makes connecting systems flexible and easy.
• STP solutions, with their demanding processing loads and complex interactions between applications and technologies, pose significant design risks - especially around performance and scalability. FSIs should use rigorous techniques, like benchmarking and predictive modeling, to help them architect systems and choose the appropriate design options. This evaluation confirmed CSC's belief in using a performance-based approach for designing and implementing high-volume systems.
Our GSTP testing demonstrates how a scalable STP solution can be built in a timely and cost-effective manner. Using proven message queuing products along with the appropriate design approaches, like dynamic routing, can facilitate the integration of disparate internal and external applications while minimising risk. Designing and implementing a flexible architecture can reduce the deployment, testing, and maintenance costs. Innovative ways to maintain state data and the feasibility of using transactional integrity with high-volume workloads were also demonstrated.

Philip Czachorowski is a partner in CSC's Consulting Group with responsibility for its Systems Performance Centre.

Comments: (0)

Comment on this story (membership required)

Related blogs

Create a blog about this story (membership required)

Solution source

Search by company or single key word