Analyst Perspectives

Increasingly, global financial markets compete on speed, so much so that high-speed trading capabilities have become a performance differentiator for the largest financial services firms and some investment funds. Transmitting messages with quotes, prices and trade data is a core capability for currency dealers. Informatica recently introduced Ultra Messaging, which is designed to offer global currency traders an efficient, high-throughput, lower-latency (that is, faster) and more secure method of linking their worldwide operations.

Currency trading, like much of the financial services landscape, has been transformed by IT. When I worked on a small currency trading desk in 1978, the two most sophisticated pieces of technology the traders used were the fax machine (in those days, before it became a consumer electronics item, a fax machine cost $10,000, or about $35,000 in current dollars) and a Monroe desktop financial calculator. Back then, a one-day movement of 50 “pips” (0.0050 of any currency unit or a half-cent) was considered a huge move in the market. Today, in an era when currency swings of several cents in a day are common and daily trading volumes are measured in trillions, computers and the networks that connect them are integral components of any dealer’s operation, helping to manage trades, searching for arbitrage opportunities and enabling financial institutions to carefully monitor their risk exposure to their customers. Traders, when they are involved, must have up-to-the-split-second information. Those that are charged with managing counterparty risk must be certain that they have real-time information about their global exposure to individual credits.

The worldwide currency market operates around the clock, and major trading centers (Tokyo, London and New York, for example) have normal working hours that overlap over the course of the day, so it’s important that the individual trading desks around the world have the exact same information simultaneously. If they don’t, they risk having competitors arbitraging their bids or otherwise trading against themselves.

Ultra Messaging is designed to optimize performance across wide area networks (WAN). Managing the flow of quotes and trade information is less difficult on a local area network (LAN) since bandwidth can be enormous, distances are short and messages on the network cross a limited number of nodes. When networks must span the world, though, latency develops because of the sheer distances involved, the sometimes complex routing a message takes between two points and potential bandwidth limitations in the technology that’s employed to handle the message. WANs also pose problems of lost data because of the relatively high number of nodes between the sending and receiving points.

Ultra Messaging optimizes the path across a WAN by continuously monitoring the network topography and using algorithms to select the optimal routing of messages for each specific moment, an approach that seeks to achieve an “as soon as physically possible” (ASAPP) latency. It is designed to offer graceful failover capability because it also has calculated the next best routing if that becomes necessary. Informatica claims it can regularly save tens of microseconds compared to other solutions on the market.

The software is available in three versions to address different optimization requirements. The Persistence Edition focuses on guaranteed messaging with zero-latency failover to address the need for institutions to have an accurate global record of all of its trades. The Queuing Edition is designed for guaranteed “once-and-only-once” delivery with low-latency messaging and automatic load balancing for constrained bandwidth environments. The Streaming Edition is designed to achieve the lowest possible latency by using “nothing in the middle” methods that eliminate daemons and brokers.

Global financial services companies have options when it comes to managing their WAN-based messaging and communications. They should investigate whether Ultra Messaging will improve their network performance.

Regards,

Robert Kugel – SVP Research