| Arthur Cole recently spoke with Jake Sorofman, vice | | | | centers where you can actually start shifting |
| president of marketing, rPath, regarding virtualization | | | | application workloads across clouds. When you start |
| and cloud computing. | | | | to compare the cost, service and performance |
| Cole: rPath specializes in application virtualization, | | | | advantages of moving workloads, it really starts to |
| among other things. How does this extend the | | | | look like electricity. |
| benefits of the virtual platforms already in place? | | | | Cole: So how do we get from here to there? What |
| Sorofman: Virtualization is part of the reinvention that | | | | sort of investments should enterprises be looking at |
| needs to take place in the enterprise because the | | | | going forward? |
| economics are broken. The business model no longer | | | | Sorofman: There is a series of investments to be |
| works. The notion of virtualizing applications is | | | | made to get us there. First is virtualization of |
| essentially a way to deploy applications in a virtual | | | | infrastructure and then on the application side. Level |
| fashion. Applications are run on hypervisors and | | | | two is experimentation - using Amazon's Elastic |
| moved across a computer environment very | | | | Compute Cloud as a basis for experimentation. Get |
| seamlessly. Application development folks are used to | | | | your hands wet, your feet dirty. Deploy some |
| creating working software and throwing it over the | | | | applications. Understand what the cloud looks like, and |
| wall to the operations folks. It's a laborious process | | | | adopt some best practices to be used as the basis |
| of manually pairing software with hardware, | | | | for learning. |
| optimizing it and finally certifying it. It's a process that | | | | The next step is to lay the cloud foundation. Take a |
| can take four to six months. But today's business | | | | reference architecture and codify it as a set of |
| environment requires rapid deployment, and | | | | processes and best practices that allow you to scale. |
| virtualization takes the friction out of that process. | | | | Then start releasing applications in the cloud on a |
| Cole: But aren't you describing more of a cloud | | | | limited scale, but enough that you begin to |
| environment, rather than a strictly virtual one? | | | | understand the complexity. Once you have the |
| Sorofman: Virtualization provides the underpinning for | | | | architecture in place, the next step is to connect an |
| the idea of cloud computing. The cloud adds a | | | | application pipeline to the cloud, whether it's an |
| dimension of elasticity. While the previous model | | | | external or internal cloud, and start the broad-based |
| required investment in server capacity and application | | | | deployment of applications. |
| workload management to meet peak demand, the | | | | The final step is actualization, where you have this |
| elastic computing model requires that you only pay | | | | intergalactic notion of dynamic workload provisioning |
| for what you consume. Applications can be | | | | across clouds. This is finally where you start to see |
| provisioned, scaled, descaled, made available on | | | | the real cost savings and service advantages of your |
| demand, and paid for on a utility basis. | | | | investment. But be aware that there is nothing simple |
| Cloud computing is the convergence of a few trends: | | | | about this. It will be fairly complicated, and many of |
| virtualization, SaaS, utility and grid computing. | | | | the technologies are not yet mature enough to |
| Ultimately, as these all mature and come together, | | | | support this vision. |
| the cloud becomes an interconnected web of data | | | | |