DeepField Networks, an Ann Arbor, Mich., startup, is coming out of stealth mode Tuesday to announce an analytics tool that maps the traffic patterns, application performance and cost structure of networks for cloud computing companies, content providers and carriers.
The venture-funded firm was created by veteran entrepreneurs including Craig Labovitz, a founder of Arbor Networks, which provides distributed denial-of-service (DDoS) protection services to carriers and enterprises.
Founded in fall 2011, DeepField Networks has 19 employees and a handful of initial customers. The company received $1.5 million in early-stage financing from DFJ Mercury and RPM Ventures. Asked where the startup gets its name, Labovitz explains: "Studying the insanely complex and expanding structure of the cloud/Internet across millions of sites and tens of thousands of data centers around the world often felt similar to the incredible exploration and discoveries coming from the Hubble Telescope [and its DeepField images]."
DeepField's flagship product is patent-pending software that provides a "cloud genome" of a network, including all of the complex relationships and interdependencies between a network operator's own gear, pipes and applications as well as services purchased from external providers. The software continuously maps millions of Internet services to discover the components and physical infrastructure underlying websites and services. It uses virtual machines rather than physical appliances to gather information about a network's traffic patterns, which is less expensive and easier to scale than hardware.
RELATED: 10 most powerful cloud companies
DeepField is initially offering two products, which are both in beta testing mode. The first is cloud genome mapping, which is offered as a subscription service, and the other is the cloud intelligence virtual appliance, which is sold using a licensing model.
These products "provide scalable, real-time, actionable analytics and business intelligence to cloud providers, content providers, [content delivery networks] and carriers," Labovitz says.
"It's a pretty interesting service. There's nobody in the market doing what they are," says Dan Rayburn, a principal analyst at Frost & Sullivan who has been briefed by DeepField Networks. "I like what they're offering. It's definitely something that's needed in the market. And it's definitely not a market where there are four or five other competitors."
DeepField's products help network operators discover unknown dependencies in their supply chains. For example, when Amazon's Elastic Computer Cloud suffered an outage in Northern Virginia in June, several other Internet services including Netflix, Pinterest and Instagram went down, too. A company with an application based on Pinterest may not have realized that it was dependent on Amazon ECS without a cloud genome map of its own service.
The cloud genome map and the network telemetry gathered by the virtual appliances are designed to help network operators launch new services, enhance network performance, understand and optimize their costs, and increase profitability. Together they provide visibility into a network's components, costs, usage and performance.
"The key differentiator for us is the level of visibility we provide," Labovitz says. "We don't just tell you how much Netflix is running on your network. We tell you how much it is costing you to deliver that Netflix traffic."
Rayburn says it's too early to know how big the potential market is for DeepField Networks' service. "Clearly, this is something that customers are looking for," he says. "They've got a bunch of really smart folks who have a history of building services that work really well. These guys have a track record of building good stuff."
While DeepField's initial customers are service and content providers, the startup sees potential for the enterprise market.
"The cloud is fantastic for the enterprise because anything can be done from anywhere, but the challenge to the enterprise is understanding the supply chain. A lot of companies were not Amazon ECS customers, but they had services go out in June," Labovitz explains. "It's similar to when companies started multihoming their networks, and they discovered that both carriers they used leased equipment from the same provider. It's fascinating to do the mapping and discover that everything is dependent on a small number of suppliers."
"It would have to be a really large enterprise -- a Fortune 100 or Fortune 250 company," Rayburn says. "If it's not a large enough organization, then it is not having the problem that DeepField's service handles. I do see potential for them branching out to the enterprise market, but not anytime soon."