In anInformationWeekarticle published a decade ago, a large global investment bank stated that every millisecond lost results in $100m per annum in lost opportunity[1]This article is still quoted today and is more relevant than ever.
Ultra-low latency is central to securing a trade at the desired price before it fluctuates. This is applicable to high-frequency traders (HFT), market makers or statistical arbitrage traders. Traders must be able to access a range of applications in real-time. Applications that handle tasks such as trade placement, analysis and modeling, and settlement, are the lifeblood of most firms, and the slightest latency incurred can impact the price of a security.
A Wall Street and Technologymagazine webcast polled attendees with the following question: “What is the greatest limitation or challenge in your current infrastructure and organization around processing and analyzing real-time market data?” 43.1% stated thatlatency was their biggest concern.[2]
In networking, latency is the time it takes to move data from one network device to another, and in the world of the Internet it is measured in round-trip time. There are several factors that can contribute to latency such as packet processing speed, optical digital signal processing (DSP) time, routers changing the packet headers, traffic delays, distances between sites, switching bottlenecks and overall network performance/speed.
The need to consistently reduce communications network latency for capital markets is growing, and the ubiquity of algorithmic trading inherently necessitates firms to act on market events faster than their competition to drive profitability.
Speed is vital to traders due to the fundamental volatility of many financial instruments. Being faster than other traders is also of great significance because it can create opportunities to profit by enabling a quick response to news and other market influencing factors. This reality is fueling a race to deploy cutting-edge technology to gain an edge by reducing latency. This has advanced the millisecond environment significantly, where algorithms reply to each other 100 times faster than the blink of an eye.
The need for latency reduction is accelerating at a pace where industry competition is being defined by how quickly transactions can happen, over the transactions themselves.
In finance, business results are measured by the bottom line. Contributing to the bottom line is the number of transactions processed per minute (TPM) or transactions processed per second (TPS). Every single financial transaction such as a bank card authorization or a stock trade must process in real-time with virtually zero latency, and error free. Global investment banks, stock exchanges, traders and hedge funds stand to gain the most from real-time market data and processing of that information.
The hurdles to real-time (virtually zero latency) data include:
?Traditional store and then process systems
?Processing a combination of live and historical data in tandem
?The inability to accurately predict and rely on bandwidth in the wide area
?Managing imperfections within a stream
?Load distribution / load balancing and hardware performance
?Decimalization in real time
?Non-deterministic networks (such as Ethernet)
?Homegrown and/or poorly performing infrastructures[3]
?There are different approaches and solutions to the above challenges.
Large trading firms have gone to great expense to curb latency as much as possible. These efforts often rely on large capital expenditure on hardware and migrating to costly proprietary technologies that minimize the virtual (and often physical) distance that data has to travel. This is certainly one path to latency reduction and subsequent benefits. However, throwing endless capital at a problem is not always an option (and even when it is – it is not likely the most prudent path.)
The alternate pathto meeting the growing demands of the financial sector for low-latency connectivity centers on a vendor-agnostic solution that iscompatible and seamlessly integrates with standards-based protocols.This route cancut up to two milliseconds off the network latency.
WDM network equipment manufacturers offer equipment upgrades that deliver low latency in a compact size, with low power consumption, intuitive operation, and flexibility. These solutions do not require replacing existing networks and are simple to set up, deploy and manage, and can reduce up to 30% latency.
More traders realize this need and are now purchasing their own dark fiber, lighting it up at their own pace with their own equipment, instead of buying into carrier networks. This enables a pay-as-you-grow architecture and faster control and maintenance over the network. Since financial transactions contain very sensitive data, the equipment must have point-to-point Layer-1 security to ensure safe transfer of all information over the fiber, such as mission-critical communication and personally identifiable information (PII). Implementing security at the physical layer, as opposed to Layer-2 or Layer-3, maintains the full throughput of the data and significantly reduces the amount of latency across a network.
It is critical to improve transaction speed by reducing end-to-end latency while also improving scale and enhancing security and resiliency. These are ideal for mission-critical applications that demand very low latency, like those operating on protocols such as the Financial Information eXchange (FIX).
A scalable solution offers low latency, transparency, and high throughput while meeting today’s business continuity requirements of services such as, GbE/10/40/100Gb Ethernet and 1/2/4/8/10/16/32G Fibre Channel mix, delivering sub-microsecond latency. Pay-as-you-grow DWDM transponder solutions deliver end-to-end transparency of data, with sub-microsecond latency, suitable for synchronous backup and trading applications for distances of up to 140 km over dark fiber.
Keeping up with latency reduction to retain competitive advantage in stock trading is a marathon, not a sprint. To invest in network architecture that might provide the lowest amount of latency now, but cannot be upgraded and improved upon, easily sets up failure in the future. Its important that networks remain agile, upgradeable and scalable to be able to adjust to the rapid growth of the industry.
Koby Reshef is CEO of PacketLight Networks
