The Time for Cloud

The relentless demands of financial market trading have spurred remarkable technological innovations.

Today, trading servers are often co-located into specialised data centres managed by exchanges, equipped with specialised networking hardware and protocols. High frequency trading can call for extreme cooling solutions, and programmable hardware called Field Programmable Gate Arrays (FPGAs). While the speed of data transmission through optical fibres is impressive, microwave network links are increasingly deployed to achieve even faster transmission speeds, leveraging the superior velocity of light through air compared to glass fibres.

Meanwhile, a very different revolution has taken place in the broader tech industry: cloud computing. This revolution has had a significant impact on how software services are built and operated in most industries. The financial sector was a late adopter of cloud computing due to concerns about security and regulation, but these have been addressed. As a result, even the largest financial institutions have undergone significant cloud transformations. However trading is an area where cloud has not had the same traction.

Much of the innovation in trading technology appears to conflict with cloud, such as the push for ever-lower latency, and the use of custom hardware. The requirement for exchanges to give fair access to the market has led to practices that are challenging to replicate in the cloud. For instance, equal length network cables are ensured from all participants by co-location facilities, so that no latency advantage is given to servers sitting right next to the exchange matching engine compared to servers in a far corner of the room. This level of precision contrasts sharply with the variable latency experienced within public cloud providers.

What is the attraction of cloud anyway?

The benefits of the public cloud go far beyond outsourcing of hardware and hosting. Cloud-native architectures promise improved flexibility, reliability, security and cost efficiency. Services can automatically scale up to meet surges in demand and shrink capacity to save costs when demand drops. However, the most significant benefits of cloud workloads are the modern operating practices that accompany them. Ideas from the DevOps movement have flourished in cloud computing, supported by the ability to treat infrastructure as code. This approach can dramatically increase the rate of change while enhancing safety and quality of service.

Growing interest in big data analytics, machine learning and now generative AI makes cloud adoption even more appealing today. While these emerging technologies are not exclusive to the cloud, they are more easily accessible and benefit greatly from the elasticity and scale that the cloud provides.

The tide is turning

We are at a very interesting moment in trading technology as the movement to the cloud gains momentum. Early adoption has focused on applications with less need for low latency requirements, such as big data or post trade systems, or crypto. There are many more applications within trading that do not require the same latency as high-frequency trading systems, and we are still only scratching the surface. Exchanges themselves are beginning to take cloud more seriously, despite their interest in co-location revenue. Nasdaq, pioneers of electronic trading in the 1970s and a leading technology provider to other exchanges, are now on the forefront of this movement to the cloud as they have partnered with AWS to run cloud-enabled markets. Additionally, CME has announced a new facility being built with Google Cloud to host their futures and options markets. Other exchanges have improved access to cloud-hosted participants, or are working towards it. AWS has introduced features to better support trading workloads, such as multicast networking and progress towards deterministic fairness in networking latency.

In my own work at Iress I have seen plenty of the benefits and pitfalls of cloud adoption, which we began about seven years ago with AWS, starting with elements of our wealth management software. Migrating existing services using a "lift and shift" approach was not very compelling. However, "replatforming" to take advantage of cloud features greatly improved our reliability, security and operating model for those workloads. The best outcomes have come from building new "cloud-native" services, which can be architected to best leverage cloud resilience, performance and efficiency.

Last year, we launched a cloud native global FIX hub that challenges perceptions of what's possible for trading in the cloud with its high performance. The ability to move quickly with this technology has proven invaluable, whether it involves onboarding customers efficiently, expanding connectivity to new regions or building value added features into the product. We have embarked on a much more ambitious project to uplift the classic Iress Order System ("IOS Classic"), which has its roots in the launch of electronic trading for ASX options in 1997, with a new cloud-native execution management system. It is time for trading to reap the benefits of cloud.

This article was originally published on stockbrokers.org.au.

Public link

Please use the above public link if you want to share this noodl on another website.