"We've done some pretty deep analysis here looking at deploying on-premises or with the different cloud providers," stated Lawrence Guillory, CEO of Racemi, in this exclusive Q&A with Cloud Expo Conference Chair Jeremy Geelan. "We've found the cost to deploy an application can quickly exceed the benefits of moving it to the cloud. That's where our technology comes into play for on-boarding to the cloud."

Guillory also touches on Racemi's image-based provisioning solutions, DynaCenter software, and how enterprises are going to need the ability to seamlessly provision from their data center to the cloud and back again.

Cloud Computing Journal: What overall business problem does Racemi aim to solve?

Lawrence Guillory: Racemi's cloud migration software addresses the number one requirement from data center managers who are seeking to move workloads to private or public clouds to reduce their cost and increase flexibility. Racemi's server imaging and cloud migration technology enables enterprises to quickly capture, clone and migrate their server images anywhere within their datacenter or a private or public cloud. The Racemi automation software uses image-based provisioning to migrate server images regardless of the underlying operating system, application software, configuration, or even the physical, virtual, or cloud infrastructure. It will even translate hypervisors to ensure that virtual machines work in the new cloud environment, such as converting VMware virtual machines to Hyper-V or Xen virtual machines.

Cloud Computing Journal: And how exactly do Racemi's image-based provisioning solutions improve how data centers are deployed and managed?

Guillory: With image-based provisioning, an image of the entire server stack is copied and deployed onto the target system, and then the target system is simply booted up with OS, configurations and applications intact. Racemi automatically handles network and storage configurations on-the-fly and injects the necessary drivers, so the image works in the new environment without manual intervention. Customers utilizing Racemi's image-based provisioning can bring up more systems faster and use much less infrastructure in the process.

Cloud Computing Journal: Why is this approach better than script-based provisioning?

Guillory: Scripted provisioning automates the process of bringing a new workload online by replicating the tedious manual steps involved in a traditional software deployment, only replacing human intervention with scripts. Scripted provisioning does cut down on labor, but it doesn't speed up the process appreciably; it can still take hours to build a new system. And the scripts have to be written and continually kept up-to-date as well - quite a task when you take into account the number of differences and nuances in hardware device drivers, for example. Script-based provisioning is an ‘OK' solution when the volumes are low and speed isn't an issue.

Script-based provisioning works fine until you try to use it to build clouds and automate the provisioning and reprovisioning of resource pools. Under these conditions, script-based provisioning doesn't provide the speed, scale, or sophistication necessary to handle the job. The sheer numbers of machines and workloads that need to be brought up in a limited amount of time makes script-based provisioning a non-starter for the cloud computing era.

Cloud Computing Journal: So can I move my entire business system within a single snapshot - including my OS, applications, network and storage configurations?

Guillory: Absolutely, our DynaCenter software captures the entire "personality" of a physical or virtual server, including Windows, Linux and UNIX operating systems, applications and the storage and network configurations, in a single bootable image. This image has the ability to migrate in the datacenter or to a private or public cloud, while maintaining the personality and configurations of the original server.

Cloud Computing Journal: What kind of price-performance metrics are you able to share with prospective customers?

Guillory: We've done some pretty deep analysis here looking at deploying on-premises or with the different cloud providers. We've taken into account server, storage, network, power/cooling costs while noting different pricing models and security models, such as firewall, VPN. And, we modeled the cost for each for 228 hours per month of compute capacity, as well as 720 hours per month.

But, the fact remains that cloud deployments today are labor intensive, taking from one or two days for a simple application or Microsoft Exchange server to something like six weeks for a business application. We've found the cost to deploy an application can quickly exceed the benefits of moving it to the cloud. That's where our technology comes into play for on-boarding to the cloud. Not only do we mitigate the deployment cost, we also mitigate vendor lock-in to a specific cloud provider or virtualization technology, or for that matter, server supplier.

I have to stop short of giving you a specific number because the cost benefit is dependent on the specific customer implementation, but you can see we've done the modeling to apply to specific customer implementations.

Cloud Computing Journal: In the future what do you see as being the next "conquests" by virtualization and/or automation? Are there any unclimbed heights yet in the enterprise?

Guillory: Enterprises are going to need the ability to seamlessly provision from their data center to the cloud and back again, and to provision from physical machines to virtual machines and vice versa. Realizing the benefits of virtualization and cloud computing depends upon the ability to seamlessly move workloads between data centers and clouds, with highly sophisticated automation handling the differences between physical and virtual machines, various hypervisors, and different systems. All of this has to be accomplished without requiring hundreds of golden images, expensive service contracts, or customer investment in increasing amounts of valuable IT labor to supporting rising provisioning needs.

Cloud Computing Journal: What topics did Racemi cover in its session at Cloud Expo Silicon Valley? What was the takeaway for delegates?

Guillory: Our talk was all about integrating existing applications with private and public clouds. We covered the economics of on-premises, private and public clouds and the need for workload mobility among servers, virtualization technologies and cloud providers, along with common deployment scenarios like customer support, testing, rapid recovery/disaster recovery.

We wanted to leave everyone focused on building in workload mobility so they have maximum flexibility to deploy where it makes most sense, including cost, whether that is in the datacenter, a private or public cloud.