Balancing Latency vs. Cost
Initially, most firms focused on in-house development, as early adopters of low latency trading infrastructures simply did not see viable alternatives in the vendor community. As a result, only the most technology savvy, dedicated--even obsessed--players with nearly unlimited IT budgets could compete.
However, the vendor community quickly responded to this growing need for speed. Driven by the development of high performance middleware, feed handlers, tick databases, trading platforms, cost effective global connectivity options, and managed hosting services; firms entering the low latency market today have plenty of vendor options to choose from, in addition to building proprietary systems and infrastructure.
The fastest 5 percent of trading firms today are now looking for sub 100-microsecond gains, and some of the top high-frequency firms are endeavoring to shave nanoseconds from their trading infrastructures. Over the next year, nanosecond reductions in latency will increasingly become the norm in the mainstream high frequency market.
The market has reached a point of maturity where the cost of capturing additional latency gains is becoming prohibitive. Developers with the skills to take advantage of emerging technologies--such as RDMA, LDMA, IPC, FPGA to name a few--are hard to find, which means that most firms are finding it harder to keep pace in the low latency race by maintaining proprietary technologies. Equally, the race for zero latency can not be ignored, as falling behind can hamper firms' ability to compete and will introduce additional risks to their business.
At the same time, CIOs and CTOs are facing increasing pressure to control operating costs and capital expenditure, and are therefore turning to providers that can offer the right mix of low latency infrastructure and applications at the right price--driven by a need to reduce both latency and costs.