Cloud Economics: The Business Case for Cloud Computing Is Getting Harder to Ignore
As the economics of building IT infrastructure from scratch become harder to justify, Wall Street firms increasingly are looking to the clouds to lower the cost of operating and maintaining servers, networks and storage technology.
Cloud computing -- the idea of leasing infrastructure located in someone else's data center -- is so compelling because it allows firms to pay for what they consume. "In a cloud infrastructure, your resources are elastic and they can be provisioned on demand, with no single point of failure," says Prashanth Nandavanam, president and CEO of Anthrope Informatics, a consultancy in New York with experience in the capital markets, wealth management and private banking. So far, the types of applications that capital markets firms have moved to the cloud mainly include risk management programs, data simulations for algorithmic trades, historical tick data management and back testing. But while firms already are tapping the cloud for middle-office applications such as Monte Carlo simulations or accessing market data via hosted solutions, don't look for big brokers or proprietary trading shops to run their algorithmic executions in the cloud anytime soon.
Given Wall Street's emphasis on high-speed, low-latency trading, even the cloud providers are skeptical that core trading applications can soar in the clouds. "There are performance constraints and overhead associated with virtualization," explains Roji Oommen, business development manager for financial services at global data center operator Savvis, which offers Savvis Symphony, an enterprise-class cloud solution. "Where firms are concerned about nanoseconds, they're going to run their algos on dedicated infrastructure."
But even high-frequency traders can leverage the cloud. "Your smart order router and your algorithms may sit on dedicated infrastructure, but there's a constellation of surrounding applications around analytics and optimization that can run on cloud-based infrastructure, and that's what we see," adds Oommen.
"If you are a pure-play quant shop, algorithmic or high-frequency trading shop, you probably are going to be keen on having a share-nothing infrastructure," Anthrope's Nandavanam agrees, noting that for competitive reasons, quants don't want their strategies running on a machine with two other quant shops' algos. But there are areas where firms are willing to share infrastructure, he says, such as a market data feed-handler. "These commoditized components, people will be quite willing to share," Nandavanam says.
Despite the buzz about provisioning servers on demand and distributing workloads across virtual servers, firms still are reluctant to move electronic trading into the cloud, confirms Kevin McPartland, a senior analyst with Tabb Group. "At the high level, the actual sending of orders is not coming from the cloud," he says. "Today most of the orders are sent from high- frequency trading firms -- from the big banks and brokers -- and they would never use a shared infrastructure for their orders."
Fine Line Between Colocation and Cloud
Whether they want to admit it or not, high-frequency trading firms already are using the cloud, some experts suggest, by colocating their applications inside third-party data centers to be closer to the exchange matching engines. "We view them as part of the same evolutionary path," says Savvis' Oommen.
In the beginning, he explains, firms insisted on housing their applications in proprietary data centers, so large brokers and banks would build facilities in the Midwest, where they could gain access to cheap power. Today, most firms, except the largest players, no longer have an interest in building data centers, Oommen says. Because the shared data centers offer large economies of scale -- and proximity to matching engines -- there's a huge push to move applications to third-party facilities, he adds.
Neutral data centers, like those operated by Savvis and Equinix, offer colocation services to brokers, hedge funds, ECNs and exchanges, and they are turning into virtual trading ecosystems. Meanwhile, these same data centers also offer cloud-based computing services. According to John Knuff, director of business development for Equinix, the global data center operator has relationships with 200 cloud-service providers, including networking and storage providers.
Recently, Equinix has been working with New York-based Voxel, which provides financial firms with a private, high-speed cloud infrastructure that bundles connectivity, storage and compute services. And in mid-April, Equinix teamed with cloud provider Logicworks, whose infiniCloud infrastructure allows organizations to migrate mission-critical applications to secure, compliant private clouds or on-demand public clouds.
Despite the boom in cloud service providers, however, "There's no rush to be an early adopter in the space," admits Knuff. While there are a lot of private cloud platforms being built globally, firms tend to be conservative with their implementations, he says, noting that it took time for Wall Street firms to accept colocation, too.
Ivy is Editor-at-Large for Advanced Trading and Wall Street & Technology. Ivy is responsible for writing in-depth feature articles, daily blogs and news articles with a focus on automated trading in the capital markets. As an industry expert, Ivy has reported on a myriad ... View Full Bio
Tweet This