How Telcos Could Conquer the Cloud

Telecom providers could play an important and lucrative role in the burgeoning world of cloud computing by combining their natural advantages as network operators with a new wave of innovation. But as they are already behind in this race, they'll need to move much faster than their current approach to technological evolution allows.

The opportunity represented by cloud-based services is potentially immense because, for starters, it increases the value of carrier networks in multiple ways and creates new roles (and revenues) for service providers. At minimum, clouds will greatly increase network traffic and utilization and thus transport revenues. And in physically delivering cloud-based services, telecom carriers have an opportunity to extract two revenue streams from a similar function: charging end users for a given level of service quality and, at the other end, charging cloud service providers for the same thing — an arrangement like that often discussed in the context of content delivery networks.

However, telcos could do much more than that thanks to some inherent advantages, said Tom Nolle, president of CIMI, a consultancy. Their IP infrastructure could lend itself well to cloud services — at least compared to enterprise infrastructure. The highly componentized nature of telco operations software gels well with cloud services because each software component can be treated like a hosted application, which yields parallels in processing and performance, Nolle said. Meanwhile, emerging telco technologies such as IP multimedia subsystem (IMS) and next-generation network (NGN) service architectures are developing into logical service components that also could fit comfortably in cloud environments.

“A service provider using cloud computing could build a supply-side architecture that matched up with the current application and service-logic trends emerging independently,” Nolle said. “It's like a combination of two perfect storms.

However, to play a bigger role in the cloud, telecom service providers will need better technology than the stuff they settle for today.

Router oligarchs Cisco and Juniper Networks both made big splashes this year by announcing plans for data center evolution that would help usher in the cloud services era. Juniper — having previously offloaded its routers' control planes to discrete servers and, more recently, having added virtualization capabilities in its networking gear that echo the same trend in servers — announced in February it was working on a “flat, lossless” data center fabric, code-named Stratus, designed to scale to thousands of 10 Gb/s Ethernet ports and drastically reduce latency in data center networks. The next month, Cisco — to much greater fanfare — announced plans to integrate networking, storage and computing in a common platform. Juniper didn't say when Stratus would be ready, though analysts have predicted its emergence late next year. Cisco expects to start shipping by June.

Still, several sources said telecom equipment vendors are lagging behind in the race to drive cloud-enabling technology.

“The vendors are all kind of at first base at best,” said Dave Schaeffer, CEO of Cogent Communications. ”They're very embryonic in this.”

Thus far, telecom equipment vendors have focused too much on the mere connectivity aspects of the cloud, Nolle said, and not enough on a more holistic view of its most important property: the efficiency with which resources are allocated for, and consumed by, applications.

“The network is the connection fabric that builds the cloud for cloud computing, but a bunch of servers on a network and a cloud are distinguished from one another by the effectiveness with which you can allocate the resources from the pool to support the mission of the application — that's the key ingredient,” Nolle said. “Just because I can connect to a resource in Singapore that's half the price doesn't mean that if I put it into this application, I'll get a satisfactory performance. Like in origami, gluing pieces together to create something is definitely essential to a cohesive structure, but having some notion of what the hell you're trying to build is even more essential.”

Even those most recent innovations in data center telecom equipment are weakest when it comes to the process of smart, efficient resource allocation that Nolle said is the most important aspect of cloud computing — a function he calls brokerage.

Some of the biggest names in cloud computing — such as Amazon.com and Salesforce.com — got there quickly by blazing a trail in brokerage. Amazon's Elastic Compute Cloud service — beta trialed last August and available since October — is sold in units of virtual machinery the company calls “instances,” which users can easily and dynamically turn up and down, paying only for what they consume. Amazon uniquely divides clouds into regions and availability zones, and it employs a concept called “elastic IP,” which sits in the gap between static and dynamic IP. Meanwhile, Salesforce.com developed a proprietary system so efficient that it is able to serve more than 55,000 enterprise customers — 1.5 million individuals — using only 1000 servers. (And since it is mirrored, the true usage is closer to 500 servers, or one per every 110 enterprises.) Both Amazon and Salesforce.com achieved these results with essentially no more advantage over telecom service providers than their own ambition and aggressiveness.

“A telco looking to solve the cloud computing problem makes up a long list of business goals and posts his bill on the door of the cathedral,” Nolle said. “Then he waits for vendors to come up and read it and give him the answer to his prayers. Amazon's on the streets preaching. They're solving the problem, not asking someone else to solve it for them. Amazon has broken ground [in cloud computing] with every blessed step they've taken. And every step they took is a step the [telecom] service providers could have taken.”

Some would say carriers could begin to one-up players like Amazon by using the service quality control their networks provide. For small businesses using Amazon's cloud services, the quality of the service is determined in large part by that business' own broadband connection, over which Amazon has no control. In that sense, Amazon's is a more “ad hoc, fly-by-night” approach, said Andy Ingram, vice president of product marketing and business development for Juniper's data center business. “On a more permanent basis, I can do something like MPLS or VPLS to provide a more secure, more traffic-managed connection between the data center elements. The challenge for network service providers is: Can they move up the value chain?”

However, some say the question is not whether they can but whether they should.

“It's not clear to me that moving up the stack is a better place to be,” said Cogent's Schaeffer. “If you're a network operator, your inherent competitive advantage lies in your network, not necessarily in the guys in white lab coats running applications.“