rPath Inc., a cloud software vendor made up of a number of former Red Hat people, has announced its rPath Cloud Engine, a newly renamed version of a product that is intended to make it easier for IT administrators to install and update applications on virtual machines in the cloud, both public and private. The product, based on the company’s former rPath X6 software, doesn’t
use scripts to install applications but instead generates the software configurations itself, based on an application’s current run-time environment.
"It is interesting from a couple of perspectives," says Jay Lyman, senior analyst of enterprise software for The 451 Group. "First, it represents the next evolution of rPath's model-driven approach to simplify and eventually speed provisioning and updates. Second, the ability to generate software configurations based on the actual runtime environments of applications is integral to release automation and devops, where application development and deployment come together to more effectively leverage cloud computing resources."
The rPath Cloud Engine’s new features, such as a query tool access feature, are particularly intended to make it suitable for embedding in private hybrid cloud stacks, says Shawn Edmondson, VP of product strategy for the Raleigh, N.C., company. It is a software product that runs on the back end, on either a server or a virtual machine, using a representational state transfer application programming
interface. There are two ways to deploy applications, he says. First is what he describes as the old-fashioned way, using manual labor to copy disks and run installers. However, users have discovered that that method doesn’t scale, he says. Second is by using image templates, or VM images that can be cloned for new applications, an approach that VMware encourages with its vCloud Director product. While that method enables an IT administrator to deploy systems quickly, once they’re up and running, it doesn’t help organizations with the change management problem of updating the applications, he says.
Instead of automating directions using functionality such as scripts or runbook automation, the software uses an approach that takes a model of the system’s current state, and then looks at a model of where the user wants the software to be. From there it creates and carries out the models, as well as provides a user interface, says CTO Brett Adam. From that, it can generate images and provide automatic upgrades. In addition, it uses a factory-like automation server and provides application programming interfaces that can be integrated in other vendor products, such as cloud management technology, he says.
The software can be sold through a number of methods, including by subscription and by a perpetual license, and is priced in a number of ways, such as by virtual machine or by blade. A typical enterprise spends $250,000 to $500,000 per year, Edmondson says. It supports the major virtualization formats and hypervisors. Users can buy it to create their own stacks, or go to a vendor and get the entire stack in one step, Adam says.
See more on this topic by subscribing to Network Computing Pro Reports Research: Virtualization Management (subscription required).