The Abiquo vision represents a high level view of the future of Cloud Management.
Our vision is founded on a clear strategy, rather than tactically crafted to fit the implementation of an existing product, or the nomenclature of confused terms.
Virtualization was a great advance in computing technology. Initially deployed as a development tool to assist efficient testing, and to reduce the need to replicate hardware, over the last decade it has become widely adopted in enterprise datacenters and hosting environments. While we theoretically enjoy the considerable benefits that virtualization promises; improved utilization and efficiency, resilience, scalability, portability and energy saving, the reality is that virtualization has simply added to the complexity, with many IT organizations finding it increasingly difficult to "keep their hands around the infrastructure". The prospect of adding to this confusion with outsourced infrastructure (always assuming the fundamental issues of security are resolved) fills many IT managers with dread. Fundamentally, management of corporate IT resources has not changed in decades, if ever. Virtualization is merely a tool in the IT organization's arsenal. Again, it's theoretically easier to provision a virtual server than a physical one, but in reality the benefits are often negated by the "virtual sprawl".
In the Abiquo vision, the provision of physical infrastructure is completely separated from the virtual application infrastructure by a "Resource Cloud". Physical infrastructure, managed by the IT infrastructure organization, contributes resources to the Resource Cloud, while virtual enterprises (containing virtual datacenters, virtual machines and virtual appliances) consume it. Today, we might think of the resources as CPU cores, memory, storage and connectivity, but these terms will change to more universal units of resource as the market evolves.
The first fundamental change is that the IT infrastructure organization delegates management of the virtual enterprises. It simply creates a virtual enterprise, assigns someone to administer it, and sets limits as to the resource that may be consumed. All of this takes less than a minute. The assigned administrator could be a development lab manager, or similar, but in many cases, especially in enterprise organizations, he or she will be an IT organization professional in another area – for example the messaging group that manages operation of corporate email.
The second fundamental change is that the decision as to where a virtual machine is deployed (ie the actual physical server it will run on) is determined entirely automatically, according to policy. In fact, neither the virtual enterprise administrator, nor any other consumer of the Resource Cloud knows either. And so long as deployment has occurred according to policy, they don't care.
Policy is however vitally important. It governs obvious rules like "don't try to run this workload on a server that has insufficient capacity or the wrong hypervisor", through security policy like "this application must be network separated from that application", to "this application is so sensitive it can only run in a local datacenter". It also covers how available physical resources are to be utilized, for example "spread the load across all available machines" where maximum performance is required to "load each physical machine fully before loading the next one" in a hosting environment.
Physical resources can be provided by a local datacenter, remote data centers owned by the organization (together in a Private Cloud) or by third party providers (Public Cloud). The IT infrastructure organization has full control of the resources added to the Resource Cloud and, especially in the case of Public Cloud resources, such as Amazon EC2, of the amount that may be consumed with each vendor. Combined with workload policy, this allows third party resources to be consumed in safety, and in line with suitable security, taking advantage of standards as they evolve.
Where permitted by role, users can capture and store virtual machine templates in private, shared or even public libraries. They can combine sets of VM templates into a single appliance for easy deployment and re-deployment. Shared libraries allow the IT organization to define standard VM templates, for example built to company anti-virus, directory and control requirements. Public VM templates from reputable vendors can be downloaded for rapid deployment of complex systems, dramatically reducing implementation and evaluation times.
The Abiquo vision substantially reduces the load on the IT infrastructure organization, by delegating responsibility for virtual enterprise management. Given that virtual enterprise users don't have direct access to physical machines, and that virtual enterprises cannot exceed allocation resources (because workload management doesn't allow it), there is no danger in this delegation. The IT infrastructure organization can rest easy, focus on meeting service levels, and take a more strategic, rather than tactical, approach. Virtual enterprise administrators are empowered to create their own virtual datacenters and to assign appropriate roles and responsibilities to other users to run them. This dramatically reduces the time taken to provision new servers and deploy new applications, allowing short term projects and flexibility that is unrealistic in current IT environments.
Finally, organizations of all sizes benefit from increased efficiency, business agility, and lower costs.