March 28, 2010

"We are a heavy user of virtualization technology," says Ben Stein, EVP of technology at optionsXpress. "The vast majority of our order processing and order routing engines run in a computer farm of several dozen computer servers and virtualizations."

For connectivity, the Chicago-based online brokerage firm uses Internet hubs managed by Equinix, a provider of network-neutral data centers and Internet exchange services. But while Stein is comfortable with distributed computing, he remains skeptical about the concept of cloud computing in general.

The idea of a private cloud, he notes, often is difficult to differentiate from server virtualization. In fact, "Part of what I see in cloud computing is full virtualization, where storage, network and workload are all virtualized," Stein says. "That is where you get convergence with the private cloud, and we have been doing that for over two years."

To achieve full cloud computing, however, firms must overcome the challenge of breaking down workloads to the point at which they can be rebuilt automatically so that they can be processed by any worker or any computer in the cloud, Stein asserts. "But the challenge is doing that it in a latency-sensitive manner," he adds.

WS&T February 2010 Digital Issue

"With a public cloud, you push [the workload] to the cloud and let the cloud take care of it," Stein continues. "But that's not compatible with time-sensitive workloads. The public cloud is a long way from being appropriate for financial services."

Public vs. Private Clouds

Like most capital markets professionals, Stein notes that security concerns also are a big obstacle to widespread adoption of cloud computing in the industry, although he acknowledges that private cloud providers often are able to assuage these fears. "The only form of cloud we've evaluated is a private cloud," he comments. "Without further advances, the public cloud is too much of a security risk to consider it."

Nonetheless, the private cloud still raises concerns that have not yet been satisfied for critical order flow and processing, Stein notes. "We have a double-edged sword with a lot of latency-sensitive applications and PII [personally identifiable information], and we need to know that when an order is processed, it will be done so as expeditiously as possible," he explains.

While optionsXpress leverages a private cloud of sorts for order processing, Stein points out that the unit of granularity in the brokerage's cloud is the server, not the transaction itself. "The barrier to decreasing the size of the granule is the ability to do it with all the controls in place at the server," he says.

According to Stein, one of the biggest draws of the private cloud is its ability to encompass multiple geographic areas, and he says he will continue to watch the technology as it evolves. "The portability of transactions and the ability to process our workflow anywhere is intriguing," he admits.

ABOUT THE AUTHOR
Melanie Rodier has worked as a print and broadcast journalist for over 10 years, covering business and finance, general news, and film trade news. Prior to joining Wall Street & Technology in ...