We must all reduce costs! (but at what cost?)

We must all reduce costs! (but at what cost?)

 
The Trading Mesh

 
 
By Lee Staines
Another week, another tier one bank announcing large redundancies. Another day, another article about how client service is king and therefore must be improved, but at a lower price point. Another meeting, another demand for technology usage to be more efficient while total cost of ownership is reduced.
 
These are trying times indeed for technologists operating in the world of financial markets. IT directors are continuously pushed to reduce costs. While heads of business units, operating in an ever more competitive market, are under constant pressure to do more with less.
 
This mad dash to reduce cost wherever possible – whilst not only maintaining services levels but also improving them, all while trying to improve performance – has practically reached fever pitch in today’s trading community (and in financial services as a whole). For vendors aiming to win any new business in today’s climate, the three elements of cost reduction, better performance and improved service levels therefore need to go hand in hand.
 
Long gone are the days in automated trading where, for example, an improvement of a few microseconds was met with the question, “where do I sign?” Vendors need to realise that things just don’t work that way any more. Most client and prospect conversations these days begin with the premise that if a project isn’t self-funding, then it’s a non-starter.
 
The Challenge of Ballooning Message Volumes
It is clear that this combination of priorities – cost, performance and service – and the required balance between the three is now at the heart of most business discussions around trading technology. And nowhere is this more prevalent than in the area of market data, particularly for options market participants and their partners that require OPRA (Options Price Reporting Authority) data.
 
“Nothing new here”, you may say, as dealing with OPRA volumes has always been a challenge. Over the years, that challenge has typically been met with just ‘throwing more servers’ at the problem, b ut with firms increasingly having to balance cost against performance and service, this approach is no longer sustainable. Having worked in this industry for many years, even I have been shocked at the number of servers (typically 30-50) that firms are now deploying to manage an OPRA trading environment.
 
The TCO of these implementations goes way beyond the server costs of course. The rack space, connectivity, and data centre footprint costs generally, can all add up to millions of dollars, just to maintain an environment that is far from optimal in the first place. Add to this scenario a combination of market-driven events over the coming months that will likely double this footprint and more (in some cases, firms are talking about deploying 120 servers by the end of the year, just to handle OPRA message volumes) and the challenge of balancing cost, performance and service can look pretty bleak.
 
A Different Approach
Fortunately there is an alternative that will allow firms to reduce costs in this area, but not at all cost. The answer lies in taking a hybrid approach to managing OPRA data feeds, which includes both accelerated hardware and optimised software.
 
The debate around hardware vs software – and the applicability of how a ‘best of both worlds’ scenario can be the best solution – is no better proven than in this use case. A hardware-based OPRA ticker plant can reduce a firm’s server footprint materially – potentially by one or two orders of magnitude – with just one FPGA-enabled server able to consume, normalise and distribute the whole OPRA feed, allowing the software to focus on what it does best in terms of running the trading applications, without requiring multiple servers to handle the data throughput.
 
By taking this approach, all three elements of reduced cost, enhanced performance and better service levels get a tick in the box. The requirement that ‘a project needs to be self funding’ immediately satisfies the cost argument. Add to this the exceptional speed, throughput, determinism and uptime levels that an FPGA-based solution offers and the second (performance) box is ticked. Finally, with the maturity and acceptance of FPGAs in our industry, experienced vendors such as Celoxica can ensure that service levels are guaranteed.
 
Read the original post on The Trading Mesh.

Share this:
Email this to someoneShare on LinkedInTweet about this on TwitterShare on Facebook