Today's cloud computing providers come from a variety of industries. Companies like Google and Amazon have adjacent businesses (search, e-commerce) and expertise in running computing at scale that fit well with selling IT services. Here are some of the major providers:
- Software as a service (SaaS) providers like Salesforce.com and Intuit see their cloud offerings as a way of letting users customize their core applications.
- Then there are managed hosting companies like Rackspace, Joyent, Savvis and Servepath that now make elastic (flexible, resizable computing capacity in the cloud) versions of their offerings.
- And systems integrators and consulting firms like IBM see clouds as a way of extending the services they offer.
And yet the biggest incumbents in hosting services are the telecom carriers. AT&T, Verizon and others have data centers next to large amounts of bandwidth, as well as an existing customer base using their collocation space. But with a few exceptions, telecom service providers have failed to play in the cloud computing debate.
So what should carriers do to build the right cloud offering? Of course, they need to address the usual concerns around security, compliance, governance, uptime and availability. But those are just "table stakes" for IT services.
To succeed, carriers need to make more fundamental shifts in how they do business.
In the cloud, telecom carriers' key is selling IT services, not just servers
Carriers sell two things: vertical units of rack space, and megabits per second of bandwidth. That has to change.
Cloud computing is about composed designs stitched together from several services. Amazon, for example, has done a great job of building a library of services that include: storage, message queuing, authentication, payment and more. If carriers want to compete with other cloud providers, then they need to offer a similar menu when selling IT services.
There are dozens of software projects atop which to build such services. For example, smart carriers might create a key-value storage service based on high-performance databases including Apache's Cassandra Project, its scalable distributed database, MongoDB or Amazon's SimpleDB, then offer it as part of their cloud.
Free is a cloud computing service pricing strategy
Clouds encourage experimentation because the initial cost is low. Some, such as Google's App Engine, are free for low-traffic applications. By contrast, carriers favor long-term contracts and minimum pricing tiers.
Making the change to pay-as-you-go, free-to-try computing might give the carrier's sales force ulcers, but it's a necessary shift because the more IT-oriented cloud computing providers have already set enterprise market expectations. So to compete, telecom service providers need to make it easy to experiment.
This is the strategy that gave Amazon Web Services thousands of loyal developers, and it worked partly because enterprise IT employees didn't need approval to start using the system; It was as easy as charging it to their credit cards initially.
Until carriers take deployment out of the hands of contract negotiators and into the hands of developers, they face an uphill battle.
Hint for telecom cloud computing providers: Brag about your loading bay
Carriers with managed hosting have a key advantage over pure-cloud computing providers: They'll let the customer walk into the data center (essentially their "loading bay," or the point where the warehouse meets the outside world, in warehouse business terms). In a hybrid managed hosting/cloud data center, each customer can have dedicated servers, plus access to a shared pool of elastic capacity.
This matters because, in a 2003 research paper, Microsoft's Jim Gray proved that compared to the cost of moving bytes around, everything else about computing is free. In carrier's hybrid hosting/cloud centers, data rates between colocation space and the cloud can be much higher, making it possible to "burst" into shared infrastructure as needed.
Carriers' major advantage over cloud computing providers: Owning the network
Telecommunications companies have access to many networks through peering arrangements and connections to local carriers. This means they can decide what computing happens where -- something pure cloud computing providers can't easily do because they rely on a degree of abstraction to separate virtual computing from the underlying computers.
As application developers get increasingly concerned about latency and user experience, this is a huge advantage. Carriers can give their customers much more control over geography and network capacity than other cloud contenders, and should build self-service provisioning of the network into their offerings.
The race to claim the cloud is just getting started, and it'll be a long, slow migration from on-premise to on-demand computing. Carriers can win that race, if they're willing to shed the baggage of how they do business today.
About the author: Alistair Croll is a principal analyst for Bitcurrent, where he focuses on Internet technologies, product management and the Web. He coordinates many technical events, such as Interop and Cloud Connect, and has authored Managing Bandwidth: Deploying QoS Across Enterprise Networks (Prentice Hall PTR, 1999) and co-authored Complete Web Monitoring (O'Reilly) with Sean Power. He previously co-founded Coradiant -- which grew from Networkshop -- where he managed product strategy, engineering and marketing.
This was first published in February 2011