Businesses should be formulating cloud strategies now that get the most out of their network providers to better support whatever cloud services they wind up buying, according to a prominent cloud-economics expert.
Businesses should be formulating strategies now that get the most out of their network providers to better support whatever cloud services they wind up buying, according to a prominent cloud-economics expert.
Latency, quality of service, and bandwidth on demand are some of the service characteristics businesses may already buy for their WAN connections that they will want to leverage as they move toward public cloud services, says Joe Weinman, founder of cloudonomics.com and a worldwide segment executive of HP's media and entertainment business.
EXPERT GUIDE: Cloud management software
"Enterprises can leverage their existing private networks with the emerging set of service provider offers," says Weinman, whose book "Cloudonomics" is due out this spring.
Services that businesses already use to connect data centers, for example, could be economically expanded to connect to cloud storage services as well, reducing the threat that performance will deteriorate when cloud services are switched on, he says.
Technologies such as MPLS, VPLS and DWDM are available and can be called into service for cloud connectivity, he says.
Doing the math on cloud economics is more than simply whether the bottom line is less, he says. "When people talk about the business benefits of cloud the first thing that they always talk about is cost reduction, and that leads you inevitably to a hybrid architecture of enterprise data center with service-provider cloud resources being economically optimal," Weinman says.
Through writings on his website and in his upcoming book, Weinman demonstrates that cloud services make such good sense economically that hybrid cloud networks for businesses are inevitable.
Businesses that have steady demand for applications and data and also have in-house expertise may as well build their own clouds. But for those who lack expertise or whose demand spikes periodically may want to go with pay-per-use cloud services.
At first blush, pay-per-use prices may seem more expensive than owning the infrastructure, but if capital costs, idle capacity that is used infrequently and maintenance are factored in, cloud services as a supplement to owned infrastructure make sense.
"If you have variability [in demand], it turns out that even if the service provider is much more expensive than do-it-yourself ... unpredictability drives you to leveraging on demand pay-per-use, a.k.a. cloud resources," Weinman says. An example would be a retailer that 11 months out of the year uses its own data center but then for one month needs to double or quadruple capacity.
If these services can be accessed via network services that can be bought on a bandwidth-on-demand basis, the economics improve, he says. Businesses can get the network performance they need without having to overprovision bandwidth for everyday use.
Most businesses have a combination of needs, with one group of applications receiving predictable, steady demand while others spike unpredictably or seasonally, leading to Weinman's conclusion that hybrid clouds will proliferate.
Service providers have some work to do in order to support this kind of service tailored for hybrid clouds, he says, despite the fact that appropriate technologies exist. But carriers have to have the back-end tools for billing customer for the services they use. "It's not helpful to have the flexibility but still be charged for the whole amount regardless of what you use," he says. Self-provisioning portals are also needed, he says.
Carriers will respond to this demand, he says, because they themselves need to manage bandwidth. Customers will increasingly want seemingly unlimited bandwidth for more demanding applications, and providers will need to intelligently and dynamically allocate it in order to make the best use of their networks' capacity. "What's required obviously is immense bandwidth, but also in the most cost-effective manner rather than giving the best possible treatment to all data," Weinman says.
"There are a lot of different things you can do that are a mix of endpoint, network and consolidated cloud services. It's going to take intelligence rather than some mythological infinite free capacity to make all that happen. And that's basically the premise; there's a need for intelligence in networks."