RPath has added a drag-and-drop interface to its software stack building system, now called rPathX6, to make it easier to build a virtual appliance or other software stack and track each of its configurations.
In the past, rPath has been a specialist tool supplier, giving early cloud users, among other things, a system with which they could construct a workload as a set of virtual files that could be sent straight to Amazon's EC2 cloud.
Now it's branching out to offer a "service factory" for the visual assembly of software stacks, storing them in a repository and deploying them to any one of several possible virtual machine environments. The system images generated must run with either Linux or Windows, but rPath is an example of how deep expertise in operating systems translates into workload preparation for both virtualized data center and cloud environments.
A bit of background: rPath is a company that was founded by Eric Troan, the former VP of engineering at Red Hat, where he developed the package method of handling Linux modules. Components of the user's choice are often added to the kernel to make up the Linux operating system, and each package includes header information identifying a module's origin and dependencies.
At rPath, Troan & company -- several former Red Hat developers among them -- have combined knowledge of Linux with the ability to package applications into virtual appliances. For a given application, the unnecessary parts of Linux are stripped away, leaving a smaller footprint and more efficient runtime code. It can then be cast in a virtual machine format, ready to run in an open source KVM, Citrix Systems XenServer, VMware ESX Server, Microsoft Hyper-V, open source Xen, or even Oracle's Virtual Iron-based hypervisor.
That means X6 is capable of producing streamlined workloads for the following clouds: Amazon Web Services EC2, using the specialized Amazon Machine Images, a variation of Xen, Rackspace, Bluelock, open source Eucalyptus Systems, open source OpenStack, GoGrid, and Globus.
rPath itself may not be the wave of the future, but its approach to building virtual appliances is likely to become more widely adopted by IT. Virtual appliances, of course, are workloads that can be shipped off to be run in a private or public cloud, and a flexible virtual appliance builder would give IT managers choices in workload deployment. That said, so far rPath has appealed mainly to Linux users, with Windows users finding it less compelling.
The process can also be used with Windows, but Troan acknowledges that the many interlocked parts of Windows make it unwise to strip out components and the system doesn't do so.
It used to be a command line experience to accomplish this application/operating system packaging. Now, said Jake Sorofman, CMO, it's more of a drag-and-drop or push-button experience, opening up workload preparation to parties outside the experts in IT.
In effect, X6 creates a blueprint of a software system and captures the configuration in Conary, its underlying open source repository. Then users of the interface may create variations of the blueprint as a deployable software stack. For software testers, being able to rapidly compose test packages of an application with variations of an operating system and database connectivity would speed their test processes. Application deployers also like to test production systems in a target environment before deployment, and X6, with its new interface, eases the burden on them as well.
X6 will also work with open source configuration engines, such as Puppet, Chef, and Cfengine.
The future of IT lies in storing system knowledge in an automated system and putting it to use when a new system is about to be unleashed. Historically, the rPath product line has been "built by and for geeks," admits Sorofman. With its drag-and-drop interface, "we want to delegate control to a broader group of managers."
That means, instead of knowing all about Tibco Rendevous or Oracle reporting modules, you could go to a component library in X6, find the names of what you want, and add them to the model of the software stack you're building. The underlying system will pull the pieces together into a deployment package for the named data center or, for that matter, the external cloud, without the user needing to know a lot more.
I can't guarantee it works as advertised. But I know for IT to get its work done in the future, it's going to have use automated systems that work something like this one.
Charles Babcock is an editor-at-large for InformationWeek.