Low Latency and Virtualization: Not Such an Odd Couple After All

Today the industry is seeing two different races when it comes to low-latency messaging. On one side, there are approximately 20 firms participating in zero-risk strategies. For them, it is about obtaining the lowest latency messaging infrastructure to capture alpha and generate their trading performance. It is a very expensive strategy that few firms can afford as it requires constant upgrades to ensure that the firm has the absolute lowest latency possible.

On the other side are the remaining trading firms that are also concerned about low-latency performance, but do not compete directly against the high-frequency and latency arbitrage traders. These firms rely on low-latency connections to capture alpha, but it is the risk they incorporate into their trading strategies that provides them with their trading performance.

These risk-based trading strategies provide better economic value since they do not require firms to invest as much into their low-latency infrastructure as the high frequency traders (HFTs) do. Risk-based traders need to concern themselves with their “relative” latency—being faster than the other firms deploying similar trading strategies.

Relative Latency, Absolute Benefits
This difference between zero-risk and risk-based strategies opens a host of possibilities for risk-based trading firms not only to trade faster, but smarter as well. While the zero-risk strategy traders require a dedicated private messaging infrastructure to eke out every last micro- and nanosecond of latency performance, risk-based strategy traders can evaluate managed infrastructures—both dedicated and on-demand—for delivery of the requisite low-latency components.

A number of network providers, including CFN Services with the Alpha Platform, enable clients to share a private high-speed network and trading infrastructure—providing ready access to data, liquidity pools and financial applications in proximity locations. By sharing some of these components over a market data delivery system or a trade execution platform, firms are getting an infrastructure that is as good as it gets without the expense of building and operating it themselves.

Even on-demand infrastructures running systems in a secure shared virtual private cloud environment can deliver better performance than some dedicated hardware in a firm’s office.

For example, a hedge fund in Stamford, Conn. might have a trading algorithm that takes seven milliseconds to run on servers located in its office. Once the message leaves the server, it takes another two milliseconds to reach the proper stock market matching engine located in northern New Jersey.

The firm can reduce its messaging latency, and IT costs, by running the same algorithm in a virtual-hosted environment, co-located within or near the appropriate market center.

From a performance perspective, the same algorithm may take a little longer to operate on a virtualized server, say 7.2 milliseconds.  Include an additional 0.5 milliseconds that it takes to access the exchange’s servers and the hedge fund has reduced its messaging latency by approximately one millisecond—as well as lowered its capital costs by selecting a hosted environment.

Firms also can experience such performance improvements across multiple asset classes. Currently, CFN Services has deployed the Alpha Platform in New York, Chicago and London and is taking orders for its new deployment in Frankfurt, which lets firms trade futures, options and foreign exchange across the platform.

Speeding Through the Cloud
Many firms may first dip their toes into hosted cloud computing offerings through their development teams that look to spin up and retire virtual servers as needed. Yet, that trend is changing as we have seen faster adoption coming from the front office, where executives are looking to lower latency and reduce costs.

Using highly tuned servers in a cloud environment that are also housed in proximity to the important global market centers, trading firms are experiencing the best of both worlds—lower costs and reduced latency.

Looking forward, additional opportunities exist to leverage shared proximity-hosted cloud solutions.  Most firms are not using proximity-cloud-based storage to store large datasets, such as tick history databases. These continue to reside in network-attached storage located in line with the firm’s internal IT infrastructures.

However, this might change shortly as CFN Services plans to launch a new Database-as-a-Service (DBaaS) offering. The new offering will eliminate individual clients from storing similar data within the same cloud and simplify data management. Instead of hosting multiple tick history databases, CFN Services would host one to which clients could subscribe. This marks another evolutionary step for cloud computing—allowing market participants to access large datasets without having to manage or store the data themselves.

Mark Casey is president and chief executive of CFN Services, a provider of automated trading enablement services to some of the world’s most sophisticated financial markets participants. Under his leadership, CFN developed and operates the ultra-low-latency Alpha Platform, a global high-performance private cloud for automated trading on key liquidity venues in equities, options, futures and FX. Visit www.cfnservices.com for more information.

Sponsored content