The beauty of the cloud lies in the way it treats resources, information and services as a utility (like an electricity grid) shared over the internet or other networks. Storing data in the cloud, for example, is a valuable back-up, or a cheaper alternative for smaller companies than buying in heavy-duty computing resources.
The EU-funded 'Resources and services virtualisation without barriers' (Reservoir) project was set up to give cloud computing a boost, to create the software platform and protocols necessary to allow small and medium-sized enterprises to combine their data centres into a large, federated cloud service that could compete on a larger scale.
This industrial-scale project has resulted in several spin-off technologies, including the 'Virtual application networks' (VANs) developed by IBM's Research Lab in Haifa, Israel. The VANs enable the dynamic creation and migration of virtual networks that can extend beyond subnets, websites and even companies.
The VANs technology is essential to achieving Reservoir's vision of federated cloud computing, because it creates an abstraction layer that breaks the link between the logical network topology - like the map of the network - and the physical infrastructure. As such, an application can be deployed, without any modification, in a single cloud or split between different clouds.
In fact, these virtual networks are movable and location independent, so components of any application can freely move within the federated cloud without losing connectivity. It is sophisticated technology that pushes the cloud paradigm into new territory. And the IBM team is keen to further develop the VAN 'proof of concept'. Derived from Reservoir's achievements, IBM Israel is also developing technologies to provide cloud administrators with advanced automation tools and created a new 'Placement optimisation engine' (POE) and a new admission control tool.
Engine and Tools
Cloud administrators use POEs to find the best load-balancing pattern to cope with demand, but the Haifa team refined the concept much further. The new POE can accept pluggable optimisation targets, so administrators can pick and choose between priorities. For example, one administrator might want optimal load balancing to distribute workload between all physical servers, in order to maximise performance, while another might target power consumption by consolidating the load into the fewest possible servers.
What's more, the system can be designed to switch between policies over time. For instance, administrators can mandate maximum performance during business hours and low-power consumption at all other times. The POE can even optimise a particular process within the cloud based on concepts like cost, geographical location, security and trust.
The admission control component is perhaps even smarter. It uses a technique, called statistical multiplexing, that allows cloud administrators to over-commit resources and still protect the 'Service level agreements' (SLAs) that are so important to the smooth functioning of cloud services. The statistical techniques essentially minimise the probability of congestion, even when total capacity may be lower than the total demand.
These components can also be integrated into different cloud management products or platforms, such as Claudia and OpenNebula - an open source cloud management toolkit for building private, public and hybrid clouds. Thanks to Reservoir's works, the OpenNebula platform is being rolled out as an enhanced commercial version.
C12G Labs is offering consulting, training, engineering and support services for OpenNebulaPro. C12G Labs wishes to emphasise that OpenNebula is not a 'feature limited' edition of OpenNebulaPro. Rather, the Pro version has additional tools that simplify the deployment and operation of clouds. C12G Labs manages the OpenNebula.org project and contributes to its long-term sustainability by dedicating an amount of its own engineering resources to support and develop OpenNebula.
The Reservoir project's input is now considered to be an important development step for OpenNebula, because most of the platform's more advanced features have been driven by the project's use-case scenarios. OpenNebula now offers a range of advanced tools for the management and support of cloud federation and interoperability.
Project partner Telefonica I+D has created a spin-out from the project with plans to use Claudia to offer global cloud services for telephony. The platform allows service providers the chance to better manage the many different types of cloud or services currently deployed in a dynamic way.
For example, it supports 'Infrastructure as a service' (IaaS), which typically means supplying service providers with data processing power and storage resources. It provides the 'Platform as a service' (PaaS) model, too. PaaS typically supplies developers with a cloud-based environment for developing applications or services by providing useful building blocks. 'Software as a service' (SaaS) is another type of cloud configuration which hosts applications. These applications can run from simple programs like word processing right through to 'Enterprise resources planning' (ERP) tools.
And the Claudia platform worked on in Reservoir manages services as a whole, controlling the configuration of components, virtual networks and storage support. It optimises usage by dynamically scaling up (or down) services as needed.
Another spin-off from the Reservoir project was the Lattice framework, developed by University College London. Lattice is a monitoring tool that has been built to oversee large-scale, highly-dynamic IT environments. It has been used in Reservoir for monitoring so-called 'running services, virtual environments and physical hosts'. Lattice is an open source framework, which can be built into many monitoring systems.
While Reservoir helped to successfully develop a wide range of technologies, many of which are now or will soon be appearing in commercial applications, the project was not just about technology. It also helped to develop expertise among a number of the consortium members. Spain's Telefonica, Thales in France and Belgium's CETIC are planning to offer cloud computing consulting services based in large part on knowledge gained in the EU-funded project. In all, Reservoir has created a wide range of platforms and tools and helped dramatically boost cloud-computing expertise within Europe.
The Reservoir project received EUR 10.53 million (of the EUR 17.17 million total budget) research funding under the EU's Seventh Framework Programme, 'Service and software architectures, infrastructures and engineering' sub-programme.