Learning New Lessons About Ultra-Low Latency Architectures

A tech course exploring big questions about the future of trading is a reminder why continuous learning is essential in business

In electronic trading, the race to ultra-low latency is essentially a race for an informational edge, a fascinating topic that was explored in a new course from New York University’s School of Professional Studies that I had the privilege of sitting in on. What united all the participants was an interest in learning how cutting-edge technology interacts with the business of trading. It covered some fascinating topics.

Taught by industry veteran Ted Hruzd, an Adjunct Instructor at NYU and VP/Architect at Citi, the course is a 20-hour deep-dive into ULL (Ultra-Low Latency) for electronic trading. It covers many different aspects of the architectures that enable modern trading, from the hardware level of over-clocked CPUs, FPGAs, and GPUs, through to the latest software advances in machine learning.

The second class I attended was not about the components that comprise trading systems directly, but covered instead the question of measurement. (Full disclosure: Ted Hruzd spoke a lot about Corvil during the session – but this was an exposition of principles and best practices, not an advertising exercise.)

Important questions

A really rewarding part of the session was the thoughtful questions that participants raised and the fruitful discussions that followed. Some were around the mechanics of instrumentation: how do you go about measuring and quantifying the performance of a system without impacting it? It turns out that it is possible, but it can be a subtle matter with some judgements to be made – is it worth adding a few nanoseconds of latency by putting an optical splitter into the network?

Other questions uncovered important operational topics, such as the challenges involved in capturing raw network data at 40Gb/s, preserving it, making it accessible and queryable in a business context.

My favorite, however, was the question about just how far the principle of colocation can be taken: if market participants can colocate their trading strategy in the same data center as the exchange, why not on the exchange servers themselves? There are a host of current challenges to this concept, some technical and some regulatory. It is nonetheless fascinating to ponder what shape and structure markets might take in the future.

Ongoing education

I’ve written before about the importance of ongoing education in the business environment, a topic I have a keen interest in, partly because of Corvil's roots in academic research, and currently because of the fascinating interaction between technology and business that we witness in our day-to-day work. On one hand, new technologies provide new solutions to old problems – "where can I get a taxi?" On the other, unfamiliar technological capabilities trail unforeseen complexities and risks in their wake – "how do I keep intruders out of my business systems when they need to be ubiquitously connected and accessible?"

Ted Hruzd's NYU course will no doubt continue to be a fascinating and rewarding exercise for all involved. The class I attended was of moderate size, just over a dozen, allowing for a very productive level of interaction with the instructor. The participants come from a wide variety of backgrounds, some deeply technical folks, and some with greater responsibility on the business side. They represent a wide variety of careers on both the buy-side and sell-side, as well as at market centers, and independent research positions.

I thoroughly enjoyed the experience and would strongly recommend it. You can find out more about the course here.

Raymond Russell

Raymond Russell, Founder & CTO
Pico is a leading provider of technology services for the financial markets community. Pico provides a best-in-class portfolio of innovative, transparent, low-latency markets solutions coupled with an agile and expert service delivery model. Instant access to financial markets is provided via PicoNet™, a globally comprehensive network platform instrumented natively with Corvil analytics and telemetry. Clients choose Pico when they want the freedom to move fast and create an operational edge in the fast-paced world of financial markets.