Success in financial services industry has long since depended on connecting to large numbers of trading counterparties, low latency/jitter and highly secure network infrastructure to power our global financial markets. We have seen disruption by way of ever faster optical networks, GPUs and switches, microwave technology, and even the seemingly counterintuitive “speed bump”. Meanwhile some 99 percent of financial market participants rely upon legacy networks built over 20 years ago on even older technology. I am talking about IP-VPN, core routing, extranets, Layer-3, MPLS, in fact pretty much all of the public internet. All of these legacy networks are non-deterministic, slow, hard to manage, slow to deliver and most have massive security flaws. Along comes the “Software Defined Network” (SDN), a catch all for the next generation network where network controllers are managed at a software layer. Adding a software layer to a legacy network is not enough to address all these flaws. In any case, financial services firms have historically shied away from the shift towards SDN, or do not want to be first movers.
However, the advantages of a well designed and operated SDN network, minus any legacy technology, is superior in almost every way and incredibly valuable to financial market firms; fast to connect to clients, agile to move or flex, low-latency and deterministic by design, secure, and far less expensive to operate. By constructing a global SDN all Layer-2 Ethernet on top of our global, low latency DWDM network, we are able to provide the next generation of networking while still meeting the stringent speed and security requirements of financial firms.
While companies like Uber have democratized the ride-sharing industry and laid the foundation for on-demand economies, SDN can do this for financial networking. Putting control back in the hands of our clients, SDN allows for near real-time control of your network.
For business users, decisions can be taken by client through the click of a button on their keyboard. This is especially important in financial services where outside factors require the ability to quickly change or move infrastructure to capitalize on new opportunities in new markets. A great example of this is where a Broker Dealer (BD) that uses a SDN decides that a specific client order flow was increasing and they allocated more capacity to that client at the press of a button. Similarly, the same BD decides to disconnect a specific client who had failed to meet their trading requirements and the head of the desk, without IT requirement, did this.
Unlike legacy MPLS systems or extranets, a software-defined network is completely transparent to the application. As the protocol used is ubiquitous Ethernet, there is very little to no additional development required on the application side. No matter where on Earth you are connecting over a software-defined network, it looks exactly the same to the application. Taking the WAN for example, it appears to an application as if you are connecting from machine to machine across your own office floor rather than connecting opposite sides of the world. The SDN standardizes and simplifies connectivity as if the entire world began speaking the same language (network language, in this case) and working together in the same office, all driven from the desktop of the business user. This seemingly flat network architecture is especially valuable to our financial system as trading applications in every corner of the world connect to hundreds of exchanges to make billions of trades each day.
Last, and maybe most important, latency performance in a software defined network is deterministic. When building a trading application to execute precise and automated trading strategies, engineers must build in tolerances to adapt to unexpected changes in latency from the application server to the exchange.
In financial services, predictability and performance are keys. SDN enables us to build a predictive network that enables a more efficient and effective platform for financial trading applications.