Key Takeaways From “Trading Fabrics” at SIFMA

Key Takeaways From “Trading Fabrics” at SIFMA

This week was the annual SIFMA conference in NYC, focused on financial services technology trends and developments. I was fortunate enough to sit on a panel discussion called ”Building a High Performance Trading Fabric…No Compromise!” which took place on June 14, 2011.

Sitting on the panel with me were Raymond Russell, CTO at Corvil, LTD and Dave Malik, Director, Solutions Architecture at Cisco.  It was moderated by Paul Jameson, Global Director at Cisco.  It was a tremendous panel and a lively discussion.

The subject of the panel was need for global Trading Fabrics.   With increased global requirements, uncertainty over regulatory frameworks, spiraling technical complexity, and the ever increasing rate of business model changes; Trading Fabrics are required to meet the strategic business goals of trading organizations. 

I thought I’d capture some of the key takeaways in this blog post.

• Elasticity: Trading systems need to be built to handle much more flexibility and elasticity. This is being driven across 4 dimensions:

o   Geographic: trading systems are deployed all throughout the world, and traders need the ability to move apps quickly into and between regions.

o   Capacity: With the continued fragmentation of liquidity venues, increasing rates of market data and order volumes, data floods are more common, and more dangerous. Built-in fabric level caching and buffering systems that can handle overflow are critical

o   Intelligence:  with different classes of applications on a single platform (algo trading, regulatory / risk / compliance, order routing, etc…) the Fabric must be able to intelligently determine how to handle exception cases for different classes of application to meet the growing divergent needs of our complex application ecosystems

o   Visibility: with elastic fabrics it’s not enough to be able to absorb and manage bursts or to ensure application stability, organizations need to be aware of the state of the entire environment to drive real-time business decisions.   

• Globalization: When trading platforms are spread throughout the world, connecting them is only part of the story. Performance matters, too. We need techniques to be able to access this global information on demand, with local-like performance and local-like coding mechanisms.
• Integrated Already: The complexity of these multi-geography, multi-application, multi-node networks can be overwhelming, especially to a development and QA organization trying to make sure the system works. Trading fabric technology has to reduce this complexity.
• Time to Market: With all the global activity, traders need to be able to spin up a new region in a week. Traditional ways of building trading systems – for all the reasons mentioned above – make this really difficult. Next-gen trading fabrics need to let developers focus on business logic, instead of just figuring out how to plug things into other things securely and reliably.

Finally, here are a couple of memorable quotes from my fellow panelists (and I paraphrase):

Raymond Russell from Corvil, LTD: It’s not always about having the lowest latency – it’s about having consistent latency with visibility into what that latency is.

Dave Malik from Cisco: System complexity grows geometrically in terms of the number of components and geographic dispersion.

Were you there? What did you think of this panel or any of your other SIFMA discussions?