An increasing number of execution systems vendors are now offering their products as hosted services, running them from their own data centres, and taking on the job of managing them, including keeping their latency low. Fidessa has been doing this for some time. IntelligentTradingTechnology.com caught up with the company’s managed services head Justin Llewellyn-Jones to find out more about hosting in a latency-sensitive world.
Q: Your job at Fidessa is to run the Hosted Services business. What services does Fidessa host, and who typically wants to use hosted services?
A: Fidessa is a broker-neutral, global market leader offering Software-as-a-Service (SaaS) solutions for the financial sector including trading platforms, connectivity and market data for both the buy-side and the sell-side.
We have a comprehensive spectrum of services that we provide as part of our integrated hosted services business. This includes multi-asset global order management systems as well as more advanced offerings such as algorithmic trading solutions and high frequency trading platforms. Our products and services are used by a wide range of market participants including boutique financial services companies, bulge bracket firms and everyone in between.
Q: Where does Fidessa actually host its services – does it operate its own data centres, or does it work with proximity and co-location facilities.
A: Fidessa does operate its own data centres, and has done so for many years. Our facilities are located in close proximity to major financial centres in North America, Europe and Asia-Pacific. We also take advantage of co-location facilities as the need arises because proximity alone often isn’t enough to support a specific business model, such as high frequency trading. We just announced a co-location deal with the TMX Group that we are particularly proud of.
Q: How latency sensitive are your customers – is there a commonality among their latency requirements or is it a broad spectrum?
A: Latency is almost always a trade off between a number of different factors – cost, value, resiliency and risk being the most fundamental. These factors have different weightings in different trading scenarios. So, although all of our clients are latency sensitive, their tolerance level for latency very much depends upon the type of business they perform. For example, a firm performing high frequency trading will always have a lower tolerance level than a firm performing a high touch agency style business when balanced against the factors I listed previously.
Q: How has Fidessa been working to reduce latencies – what are some of the aspects that you’ve been focusing on?
A: Fidessa has developed many applications oriented to different sides of the trading business. We evaluate the business focus of each customer and provide specific technologies that are suited to their needs. We work diligently on reducing application level latency because this has a direct and immediate benefit to the business user but we also look downstream at things like network topology to ensure the connectivity set-up, for example, is best suited to their needs.
Some clients require a dedicated line from their applications to a liquidity venue while others, based upon their latency tolerance level, can be served well with a service bureau-style connection. It is our job to evaluate each customer’s specific trading needs and come up with appropriate, cost effective solutions to remove latency.
Q: What are some of the 2011 initiatives you have planned, especially regarding ongoing latency reduction?
A: Fidessa has developed an exciting new HFT trading product focused squarely around the requirements of the world’s most demanding and latency-sensitive clients. In this environment, where each microsecond counts, every aspect of the Fidessa solution – application, hardware and co-location – is given exhaustive focused attention to ensure optimised results.
We remain extremely focused on innovation and remaining one step ahead of our customers’ business requirements. We have developed specific applications for our most latency sensitive customers but we are also looking closely at the service elements of our delivery model. Making sure a customer’s telecommunications provider is the best possible match for their business needs or optimising application hardware configurations are both part of this process.
Q: What is the business imperative for ever reducing latency? Is this something around which you can offer as a premium service, or do you need to do it just to stay in the game?
A: Technology dictates that processes will continue to improve and systems will become faster which means there will always be a business imperative around reducing latency.
Customers who are extremely latency sensitive will pay a premium because their profitability depends on it. The latency requirements related to a company’s network topology, data centre and its applications are high value to the business user and the cost of reducing the latency associated with them can be well worth the premium.
Over time, and in many different geographies, data volumes, DMA and high frequency trading will continue to increase. Vendors in these categories will spend a lot of time trying to reduce their latency and until we reach a point where measurable improvements become cost prohibitive, these services will still be priced at a premium.