May 23, 2015
Infrastructure Convergence: The Choice is Yours
Modern datacenter infrastructure is set to undergo a transformation as the shift towards cloud computing and massively scalable processing enables more users, applications and workloads to access underlying resources.
The constraints of traditional software and hardware models are being loosened as cloud computing gives individuals and businesses more choice over how to acquire or deliver IT services. This is prompting many to look at alternatives to the compute, storage and networking technologies they deploy today, leading to growing interest in the concept of infrastructure convergence. But infrastructure convergence covers a multitude of sins so choosing the right version depends on the business need and the use case.
The unified infrastructure approach
Virtualization changed the rules of the game by replacing rack-mounted servers and a single application to single server model with a chassis and blade server environment, known as unified infrastructure. The modern blade and chassis backplane has evolved significantly to the point where it is possible to integrate into fabric interconnects and provide massive blade throughput. Unified infrastructures are powerful and extremely agile, but they can be expensive. Organizations need to be very clear on the relevance of the use case. For example, a chassis and blade environment is ideal for a large scale-out architec
ture that requires a vast number of resources and may need to deploy hundreds, if not thousands of racks. Blades can isolate workloads, create powerful orchestration rules and provide dynamic support for business needs.
The converged infrastructure alternative
Converged infrastructure deploys node-based units that combine server, storage and compute in a single box or appliance. This is a very effective way to use data center resources more efficiently because organizations can offload big workloads to smaller converged appliances instead of buying more expensive hardware and storage. Converged infrastructure is a tremendous scale-out solution but there are some limitations. Many appliances are limited to a specific hypervisor or won't integrate with technologies like FC/FCoE.
Hyper convergence and beyond
Hyper-convergence merges all aspects of data processing, including storage and networking, in a single appliance. Because it uses software to control all resources on top of the hypervisor and abstract the management process, hyper-convergence is completely hardware agnostic. This can deliver significant cost savings when it comes to choosing the underlying hardware. Hyper-convergence technology is designed to provide the greatest scalability, regardless of hardware or hypervisor. Intelligent convergence sof
tware can translate VMs from one hypervisor to another if a business has a second location running a different hypervisor. Organizations can deploy appliances and create a central storage control infrastructure using software-defined storage policies that provides complete visibility into all storage resources while still processing
workloads on the hyper-converged platform. Hyper-convergence is a key part of the evolution to wards a more fluid data center architecture where an entire hardware stack could be abstracted and managed from the virtual layer. The ultimate ambition is to allow data, VMs and applications to flow from on-premise data centers to
the cloud and back. Hyper-convergence will enable organizations to deliver dynamic resources to a variety of points using an assortment of hardware systems with different hypervisors. It will enable the data center to remain relevant to the organization (and its users) while providing it with the agility to retain its competitive edge.
Current rating: 3.5 (2 ratings)