Service providers or carriers own the network infrastructure that supports the Internet, and are well-positioned for cloud computing. I addressed some of the opportunities and challenges service providers face in an earlier post about carriers and clouds. Interestingly, though these businesses have a service-orientation in the front-office, they still have a network operations mindset in the back-office shaped by years of focusing on call completions, tariffs, and a corresponding revenue model. Some service providers offer infrastructure-as-a-service (IaaS) and storage-as-a-service (STaaS), for example, but it’s really only outsourced resources instead of a cloud services model.
Telecommunication service providers look at cloud service providers including IaaS suppliers such as Amazon Web Services (AWS) and the platform-as-a-service (PaaS) providers like Google App Engine, and recognize opportunities to leverage their core competencies and networks for new services. To realize these opportunities, however, service providers need to think and act differently to take advantage of cloud computing and compete in the cloud market.
Mining the Network
Telecommunication service providers are looking for new services to compete against the likes of AWS and Google. While outsourced IaaS and STaaS might generate revenue, these services are really just a new twist on the old outsourced managed services business model. The actual opportunity, however, lies in data analytics. More than analyzing connectivity and network usage, something the service providers/carriers are well-experienced in, data analytics here refers to services that examine user demographics about subscribers including information about who was where and when.
An increasing number of users are mobile, meaning not tied to any physical location. Knowledge of their movement and data access patterns opens up new opportunities for providers to deliver higher value data services working with enterprises looking to give their constituents a better total customer experience. The future is in customer intimacy and the secret is location, location, location.
Challenges in the Operations
With data services, latency has usually not been too important, but with more and more people relying on their mobile devices it increasingly does matter. Case in point as told to me by EMC Service Provider CTO Paul Davey. While in San Francisco recently, Paul logged into American Airlines to change his reservation for the flight back to the UK where he lives. Because he was a Vodafone mobile customer on the AT&T network, AT&T and Vodafone repeatedly passed him back and forth resulting in it taking about 10 minutes to make the reservation change at a cost of about 25 Pounds (about $40 $US).
The point is not only network performance and the delivery time, but the cost incurred for data services as well. Carrier capacity needs to be configured to handle peak loads which results in higher costs and pricing structures for most of the world with legacy infrastructure. Emerging markets such as China, for example, have lower costs because the real estate cost of putting up an antenna is less there than in London (UK) or New York (US).
Service providers’ networks are designed for voice, not data packets, further increasing the difficulty of handling the ever-increasing demand for data delivery to a single customer at varying locations. Service providers tried to solve the delivery problem with IMS (IP Multimedia Subsystem): an all IP-based wireless network originally designed by the wireless standards body 3rd Generation Partnership Project (3GPP) as a part of the vision for evolving mobile networks beyond GSM. IMS is intended to aid the
access of multimedia and voice applications from wireless and wireline devices. Interest in IMS, however, has waned because it is now easier to access content via other means such as object technologies with its unique identifiers and associated meta data for multimedia. Given this situation, service providers cannot just re-architect for voice alone; it’s too big and too expensive an undertaking. What service providers can do is deliver and manage short-message services (SMS), email, and voicemail services because these offerings ride on top of existing networks, which allows service providers to manage performance. They cannot, however, manage data packet traffic (e.g. Google search, streaming video) because it requires them to reconfigure network capacity—which can take upwards of 3 months or more to tune for service.
Getting to the Gold
A possible solution to what Paul experienced, being routed through Vodafone UK though he was local to American Airlines on the AT&T network while in San Francisco, would have been for AT&T to provide a virtually hosted cloud service. An instance of the American Airlines reservation application, including data, could have been placed locally so Paul would not have had to transverse the network. To put a service at the edge of a service provider’s network, such as AT&T’s, the provider needs to know Paul’s location, movement patterns, and data usage; information derived from analysis of Paul’s mobile services usage. Additionally, the provider needs virtual storage. Why? Networking is more expensive than storage; networks are optimized for peak loads, meaning a good amount of bandwidth sits idle much of the time. It is less costly to service providers—and potentially, less pricey for customers—to federate virtualized data storage and keep it at the edge, closer to customers. Service providers who leverage knowledge of customers’ locations via data analytics benefit from lower data transport as well as lower energy costs due to less data movement, including during peak times. Their customers benefit from having more ready access to data at potentially lower prices.
At some point, service providers/carriers stopped using custom ASICs on proprietary technology, and moved to commodity servers placed closer to the edge of the network and to their users, rather than in a central location. This approach to network equipment is analogous to what storage (e.g. EMC storage) and servers went through when moving from proprietary technology to x86 architecture. Virtual server and networking solutions, which move the management intelligence into a control plane separate from the hardware, are now available from vendors like VMware and Nicira. Even with virtualized servers and networks, service provider costs for network data transport are higher than the cost of storage, however. To offset network costs, virtualization needs to be applied to storage too, which will enable service providers to move data during off-peak hours closer to customers over federated storage architecture. This capability, plus knowledge of the customer through data analytics, will help drive down costs and define service providers’ role in delivering new, higher value data services.