Private cloud should, in many ways, be competitive with public cloud services -- and not just an extension of a virtualized data center. Ultimately, when it comes down to picking the right IT resources for business problems, end users will make their choices between available on-premises services and comparable solutions from outside providers.
Nelson makes the observation that by listening to pundits and reading IT media, one may conclude that "everybody's on the cloud." But that's just not the case yet, she explains. While many companies may think they have private cloud in place, they are usually missing some essential component. A true full-functioning private cloud, she says, needs to meet five criteria: It needs to be standardized, automated, self-service, shared, and based on a pay-per-use model. The bottom line, she says, is the private cloud services offered need to be competitive with similar services being offered by outside Software as a Service, Infrastructure as a Service, and Platform as a Service providers.
Nelson's remarks echo those of Presidio's Steve Kaplan, who also urges market-based pricing of enterprise private cloud services.
Nelson detailed the four struggles to getting to private cloud, and provided four recommendations to address these challenges. The four challenges:
- Field-of-Dreams thinking: "If we build it, they will come."
- Creating metrics for comparison: "Metrics often don't align with user expectations."
- Achieving ROI: "Most benefits are soft benefits."
- Figuring out what to do beyond test/dev: "What applications should go in a private cloud?"
- Connect with the business: The name of the game is agility, Nelson says, and business end-users want "resources comparable to public cloud resources." They want rapid provisioning, flexibility, temporary capacity and elasticity." In addition, just as is available from public cloud services, they want self-service and self-sufficiency.
- Treat cloud as a new service: Don't try to extend an existing data center infratructure as a private cloud, Nelson advises. Instead, launch private cloud as a fresh, new set of service. The new service may use a limited amount of infrastructure, and thus may see higher utilization rates -- an important consideration for building on the success of the effort, she adds. And when it comes to measuring that success, "don't measure success based on your other services," she advises. "Compare it to the real alternative -- public cloud. Your users will be comparing those services to public cloud." In addition, pricing needs to be set along the lines of what public clouds charge for their services.
- Create incentives: "You are the cloud provider," Nelson says, noting that as with any public cloud provider, enterprise cloud managers need to attract customers by making their services fast, easy, mobile and familiar. She urges businesses to establish a chargeback arrangement based on actual usage. If a chargeback approach is not possible, establish a "showback" structure that illustrates how much business units are spending on service use.
- Decide application fit based on cloud economics: When initiating a private cloud, don't start with large, mission-critical applications, Nelson advises. Instead, opt for short-term or varaible applications. "You don't want to be moving your entire email system into the cloud as your first project," she says. Instead, "build a small environment, and fill it up as quickly as possible, so you can show executives high utilization rates." Ultimately, Nelson adds, offering competitive cloud services may open up new revenue opportunities for the organization, and even turn IT into a profit center.
The important thing is to align user expectations for private cloud to be comparable with the services and benefits they receive from public clouds, Nelson states. And, with greater agility and faster deployments, there is also some areas in which expectations need to actually be lower than that of enterprise systems. As with public clouds, users should expect less customization and a greater probability of downtime in exchange for the faster access to resources.