"The NYSE's impending data center move is on the radar screen for all of these high-frequency guys," notes Panzica. As announced last year, NYSE is building a nearly 400,000-square-foot data center in Mahwah, N.J., and moving the equity market (NYSE Arca and NYSE) matching engines out of its downtown Manhattan and Brooklyn facilities. "They're talking to all their customers, and they have room for some of their order flow servers," according to Panzica. "[High-frequency trading firms] need to move their operations potentially up north because that's where the new matching engines will be," he adds, noting that he's been in talks with HFT firms to house their strategies in Switch and Data's North Bergen facility.
"One of the challenges is, where do you colocate in a fragmented market?" points out Mark Akass, chief technology officer at BT Global Banking and Financial Markets in London, which offers the BT Radianz Proximity Solution, a global hosting service that allows firms to colocate with major exchanges and trading venues around the world with, Akass claims, "less than 1 millisecond of latency." The high-end players, he notes, want to colocate right next to the primary exchange in which they provide liquidity.
But what do they do about the other venues? One approach is to colocate with one prime exchange and then tap a low-latency shared infrastructure -- such as the pipes offered by BT Radianz and others -- to reach the other venues.
Connecting the Data Center Dots
"One of the common misconceptions is that people pick a [single] data center" in which to house all of their trading strategies, says John Knuff, director of business development at Equinix, a leading colocation service provider. "A lot of trading firms will have their network hardware and their systems distributed in several different facilities."
"Nowadays, with fragmentation of the markets, you need your central aggregation point where you do your market sweeps, and that's where people heavily utilize Equinix," explains Knuff. "If you have systems deployed in six or eight data centers, that requires network providers. You are probably spread across 30 to 40 miles, and you have to connect these [centers]."
To improve its connectivity, Buttonwood Trading Group recently moved its high-frequency trading technology to Chicago-based global hosting provider 7Ticks, which provides direct exchange access, according to Buttonwood's Dugan. "The drive to zero latency is of paramount importance," he says, adding that Buttonwood initially relied on RTS' server farms since the technology provider operates data centers all over the world.
RTS has software interfaces with more than 100 exchanges, both to receive market data and send and receive order flow, notes Dugan. But to get to the exchanges, he explains, RTS has to establish connectivity with a high-speed data communications link via a commercial telco carrier. RTS had a relationship with 7Ticks, whose core competency is global networks, server management and monitoring, Dugan continues. "From 7Ticks, we're on the best-in-class infrastructure," he comments.
Initially Buttonwood coded its strategies and applications using RTS's software and servers and then brought its own servers into 7Ticks, explains Adam Florzak, director of business development at 7Ticks.
As the software vendor, RTS codes the feed handlers to accept the native data from the exchanges and interpret it. Meanwhile 7Ticks provides the physical infrastructure -- it gets the telco lines to exchanges, runs fiber optics to the CME, handles data on its switches and provides it to servers from which the RTS application picks up the data.
"Buttonwood decided to combine the best of both worlds," says Florzak. "They were running [the code] on RTS servers. ... Now they are running it on our network with their own servers."
Meanwhile, Dugan adds, RTS conducts ongoing analysis of Buttonwood's server setup and operating system to optimize performance. "RTS has excellent microsecond time stamps for different system components," he says. "The most important area of measurement is the application monitoring and how long it takes from price update to get to the exchange. We have that time below half a millisecond, but we're [going] down into the microseconds."
