In today's world, more servers are provisioned through virtual machines than through physical hardware. The virtual data center has become the primary platform on which workloads are deployed. Typically this platform uses different components from different vendors, each with its own language, management structures, and algorithms for optimizing performance. Yet we expect this collection of disparate platforms to work seamlessly, providing us with deterministic performance levels.
Aggregating these incongruent systems into a cohesive solution using software is the goal of the software-defined data center (SDDC). This model allows for policy-driven management that aligns resources with application demand. However, integrating all these components into a single element that's manageable from a single point with universal controls is a challenge.
Instead of waiting for a universal language to stitch everything together, one trend that is emerging in the industry is the software control plane. Intelligence is moving to the perimeter of the architecture, and resources are being commoditized. For example, the network virtualization platform VMware NSX utilizes common network components to provide bandwidth and connectivity while leveraging the hypervisor to provide controls and policy-based solutions at the virtual infrastructure level.
Moving intelligence to where it matters is also occurring in the storage stack -- for example, in VMware's Virtual SAN and PernixData's FVP. Virtual SAN provides an end-to-end solution, replacing the whole storage infrastructure with a policy-based architecture for storage performance and capacity in the hypervisor. PernixData FVP aligns with NSX by decoupling performance from capacity. It leverages the storage array to provide capacity and data services while using server resources to provide storage performance and intelligence close to the application.
What these products have in common is tight integration with the hypervisor kernel. Because the hypervisor kernel is rich with information -- including a collection of multiple tightly knit resource schedulers -- it is the perfect place to introduce policy-based management engines. The hypervisor becomes a single control plane that manages both the resource and the demand. It provides a single construct to automate instructions in a single language with a model for granular levels of quality of service for applications.
Moving storage resources directly into the compute layer allows not only for greater control, but also for more detailed insight into resource consumption and distribution. Traditional storage designs for virtualized environments rely on large catchall disk pools to satisfy incoming resources, typically providing sparse information about resource distribution. This model is relegated to the past. Moving the intelligence up the stack is the natural move.
Controlling resources as close to the application as possible provides the ability to scale out swiftly and actively. Combining the powers of software and the hypervisor structure can create an application-centric platform that allows you to control your environment and contribute to business goals.