Putting the datacentre in a box

Simplicity and complexity are the ying and yang of IT. Almost as swiftly as technology promises to simplify life for IT professionals, complexity follows in its wake. Virtualisation is a prime example of a technology that has revolutionised IT and removed a whole tier of complexity and inefficiency, but not without its own cost in terms of creating strain on other parts of the infrastructure.

The traditional storage and compute approach is reaching the end of its practical life as virtualisation and the cloud dramatically alter the IT landscape. Buying SAN or NAS arrays and servers separately is complex and costly for any organisation responsible for integrating, managing and maintaining disparate systems to deliver IT services.

Converged systems have been touted for their ability to integrate SAN or NAS storage arrays and servers in a single rack. But they add little value over traditional systems because they are simply packaged storage, networking and compute components. They do not remove complexity because they are complex to maintain, often they do not provide a hypervisor or storage choice and are costly. This is especially true for SMEs and mid-sized businesses. IT leaders also struggle with getting more out of their existing investments in the long run, often wasting capacity and resources in a dynamic environment where business requirements can sometimes change overnight.

The nirvana of removing complexity would be the ability to buy a datacentre in a box. That’s the promise of hyper-convergence: modular datacentres that can be expanded by adding blocks of compute, networking and storage all bundled together as a simple box, connected to the cloud for backup and other services that are more economical to run outside the datacentre. If converged systems are separate software and hardware components engineered to work together, then hyper-converged systems are software-defined and integrated.

Due to its ability to shrink everything into a space-saving appliance form factor, there are a number of reasons why hyper-convergence is a logical progression in the evolution of computing and the datacentre.

Software-defined storage is the starting line

The latest technology trend in the data centre is a Software-Defined Storage (SDS) approach. In most cases, IT managers do not have the luxury of replacing their entire datacentre with new hardware any time they want. They have to contend with existing infrastructure whenever they add new equipment.

SDS transforms existing storage and server hardware into a pool of resources that can be utilised together by applications. Just as VMware was successful in making the compute layer software-defined, SDS vendors are making fast progress in virtualising at the storage level. The emergence of SDS makes sense as the logical next step in the software-defined evolution.

SDS brings significant benefits by enabling organisations to use commodity hardware in their storage solutions and avoid the perils of being locked into the proprietary upgrade paths determined by typical vendors. It frees businesses up to acquire fast, cost-effective storage from a trusted source on the hardware they know and love. No one has to remain hostage to steep hardware premiums or settle for infrastructure complexity.

Paving the way to hyper-convergence

An intelligent SDS solution can instantly deliver storage resources for any application more efficiently using a combination of existing datacentre infrastructure and hyper-converged building blocks.

By harnessing the power of faster and cheaper processing power, memory and flash, these solutions avoid the inefficiencies of SAN and NAS hardware, while at the same time delivering all-flash performance. They enable organisations to pool all storage and quickly deliver virtual volumes to any application and some solutions are already available today that deliver the performance of an all-flash storage array at half the cost of traditional SAN or NAS. Customers can move virtual machines and data between storage arrays and datacentres fast, even across low-bandwidth high-latency connections.

By definition, hyper-convergence is the next step because it integrates storage, servers and networking into a single system. Hyper-convergence seeks to provide flexibility in the software and hardware infrastructure used to deploy enterprise applications, while cutting costs and improving availability.

There are two main options for deploying hyper-converged infrastructure. One is through a reference architecture specifying server, networking and storage based on SDS technology, where software controls all resources on top of the hypervisor and abstracts the management process. IT professionals would have the freedom to efficiently use the hardware they already have or are accustomed to using. The second method is to deploy an appliance, which is one step closer to making the datacentre in a box a reality. The industry is already offering appliances that pre-integrate all the compute, storage, networking and virtualisation into a single product.

The datacenter in a box is here

A turnkey appliance solves many implementation and management problems while preserving the hyper-convergence benefits of cost reduction, high availability and increased agility. The simplicity and power of hyper-converged appliances make them ideal for growing mid-sized enterprises that have all the IT service headaches of big corporations but without a large staff and budget to handle them. Appliances are easier to scale because adding capacity is as simple as adding more nodes or appliances in an industry-standard x86 form factor. They are simpler to acquire because everything is covered by a single SKU and they are easier to manage because all nodes are controlled as a single unit.

Some hyper-converged appliances come pre-built on one or two server platforms making them very easy to deploy. However, by specialising on a particular hardware, they limit configuration choice and if the customer is not familiar with the hardware, these appliances may actually require additional training and services that weren’t in the original budget.

Other appliances are available with a broad choice of leading hardware and hypervisor platforms, so that customers can use what they want rather than be forced to buy technology the vendor tells them to use–and support is provided from a single source. For example, an all-flash hyper-converged appliance could provide necessities such as a three-year global end-to-end service and support program, covering the entire appliance, including the SDS technology, hypervisor, solid-state drives (SSDs) and server hardware components.

Beyond simplifying deployment, some hyper-converged appliances vendors have made enterprise-class all-flash storage an economically viable option. They efficiently utilise deduplication and I/O acceleration technology to reduce the amount of local flash required per TB of storage. These appliances can provide the effective all-flash capacity and horsepower to support hundreds of server VMs.

They are not confined to the enterprise but are also attractive to SMEs because they deliver all-flash performance at 50 to 90 per cent lower cost than traditional SAN or competitive hyper-converged appliances.

The emergence of all-flash hyper-converged systems that work on multiple hardware platforms with multiple hypervisors now makes the lives of admins easier. The datacentre has been put in its box; businesses can benefit from the agility, efficiency and assurance hyper-converged turnkey appliances built on advanced SDS technology can offer.

David Cumberworth, VP EMEA, Atlantis Computing