It’s been said that when great changes are afoot, so are great opportunities. That certainly looks true for the cloud and the decades-long pursuit for greater openness in enterprise infrastructures.
The argument is a logical one: once enterprises move data and applications off their proprietary infrastructures, open formats are the only way to ensure compatibility with whatever systems and architectures they encounter. Granted, enterprises could always insist that their data conform to a given platform, but that would severely hamper the very flexibility that clouds are supposed to provide.
This is certainly good news for the Linux community, which is actively seeking to get in on the ground floor of enterprise cloud deployments. Red Hat, for example, just released the Enterprise Virtualization 3.0 platform that some say finally kicks the feature set of the KVM hypervisor up to VMware and Microsoft levels. The new version’s management system runs as a Java application under the JBoss Enterprise Application Platform, where it has been stacked with new features such as improved self-provisioning, local storage support and an integrated RESTful Web services API. It has already received the backing of some of the top enterprise vendors like Intel, HP and Cisco.
Meanwhile, open source alternatives are popping up elsewhere in the enterprise cloud stack; even the operating system. Piston Cloud Computing offers what it says is the first cloud OS based on the OpenStack platform. The Linux-based Piston Enterprise OS (pentOS) distribution uses the company’s Null-Tier architecture to combine compute, storage and networking on the node level, providing greater scalability at lower cost than proprietary solutions. The system can be deployed in 10 minutes and can access any OpenStack-compatible environment such as RackSpace, Dell and Amazon.
The cloud, of course, is all about services so it would help if enterprises had an open platform to integrate existing service architectures with those on the cloud. That seems to be the goal of The Open Group’s latest initiatives: a new SOA reference architecture and the Service-Oriented Cloud Computing Infrastructure Framework (SOCCI). The idea is to support a vendor-neutral common language to bridge the gaps that many services encounter when trying to coordinate with each other on the cloud. Through a shared terminology, services should be able to operate within a wider universe without having to recode and remap individual architectures.
And, since the cloud relies on network infrastructure to drive dynamic data environments, more openness on the grid may be warranted, as well. Big Switch Networks recently issued an open source version of the OpenFlow Controller as part of the Software-Defined Networking (SDN) initiative. Dubbed Floodlight, the system monitors and maintains control data from OpenFlow-compatible switches from a server-based environment, rather than directly on the switch itself. This provides for greater management centralization and more efficient distribution of network services. The system is available under an Apache 2.0 license along with Hadoop and OpenStack.
Open source proponents say that not only are today’s platforms much more flexible and feature-laden than their ancestors of the data silo age, but provide much greater reliability and developer support, as well. The internet, after all, is based largely on open-source technology so if the enterprise wants to play on that field, it needs to adjust to someone else’s turf.
