Hyperconverged Infrastructure – Hype or Reality?
Dictionary definitions of “hype” include “exaggerated, extravagant or intensive publicity / promotion”, and even, “a deception carried out for the sake of publicity”! So is this all that the increased interest and attention on Hyperconverged Infrastructure is about? Or is this emerging disruptive IT solution a reality? Converging hardware and hypervisor, and their management, into one place would most definitely be a game-changer for enterprise IT – so a “Hype” or “Reality” check certainly seems to be in order.
Hyperconverged infrastructure solutions supporting software-defined data centres and cloud computing platforms have spawned a great deal of interest over the last year. Start-ups, such as Nutanix, and Simplivity have recently signed global deals with IT giants Dell and Cisco respectively. As interesting, is the announcement of EVO:RAIL by VMware – a hyperconverged solution running vSphere, vSAN and vCentre with infrastructure provided by partners which include EMC, Supermicro, Fujitsu and Dell. With such big names clambering to get a piece of the pie, hyperconverged solutions must be the next big thing….. mustn’t they?
Hyperconverged infrastructure vendors take the concept of a software-defined data centre and offer it in a single box (an x86 server). In this way, compute, storage and network components are all managed through a single hypervisor, (often along with additional features such as compression, deduplication and data protection). Conversely, traditional infrastructures will typically amalgamate solutions from different vendors, managed by different teams, supported by different suppliers brought together under the watchful eye of complex interoperability matrices. Between the two extremes sits converged infrastructures – such as those provided by VCE. Converged infrastructures are a packaged or bundled solution (or reference architecture), normally made up of components from different vendors, but supplied by a single entity and managed as a whole (or nearly whole, depending upon the degree of configured orchestration).
In this series of blog posts, Accordant’s David Boyd looks at the hype and reality around hyperconverged infrastructures. What are the relative benefits of hyperconvergence over traditional infrastructures? How might they fit into a future-state commodity-centric vision of IT? How and why should CIOs / IT Directors operate in a bimodal fashion? – keeping the lights on while transitioning to the new world.
The notion of a software-defined data centre and hyperconverged infrastructures, (solutions where compute, storage, network and often backup and deduplication are all delivered ‘in a box’ through a single hypervisor), are disruptive to the traditional IT architect.
If there is typical traditional data centre architecture it might be based on the usual players’ technologies, Cisco networks, Checkpoint firewalls, EMC storage, Brocade SAN, compute from IBM, HP or Dell, hypervisors from VMware and data protection from Symantec. It’d be a performant solution built using proven infrastructure components from different vendors, pieced together using clever cabling solutions and HBAs and NICs of various flavours, all under a careful watchful eye of several interoperability matrices and some skilled staff.
The traditional approach, has and continues to serve the enterprise well; indeed there were few real alternatives and those that were available presented too great a risk to justify change. The choice of vendors at each infrastructure level drives competition, and completion drives down price, which is better for the customer; Yet when we take a step back we see a complex web of different products from different supply chains managed by different teams, all at different stages of their lifecycle. Changes that affect service availability are raised frequently, service or usage based costing is, at best, challenging, and when things go wrong the opportunity for finger pointing is never far away. To alleviate some of these issues powerful business units establish dedicated silos of infrastructure which then multiply, and become costly and difficult to manage. This in turn leads to complex consolidation programmes (that rarely deliver their promised outcome to expected budget) before the cycle starts again. Add to this the confusion created by internal bureaucracy and the explosion of Shadow IT and the picture looks evermore complex.
A better way of doing things was needed and first to take a step was EMC, VMware, Cisco and Intel who created the Virtual Computing Environment (VCE). VCE’s line was to provide the customer with a working solution rapidly. The fact that it was made up of different vendor’s infrastructure become less relevant because it is a ratified design and is provided by one organisation. Since the creation of VCE several other organisations (HDS, HP, Cisco, NetApp, and EMC) have followed the lead, building predefined infrastructures or pre-ratified reference architectures. These converged solutions don’t have the procurement and lifecycle issues of the traditional architectures and the interoperability issues are taken care of. Yet, they are still made up of different infrastructure components (often) from different vendors, and IT organisations have struggled to understand who owns the solutions – server, network and storage teams are all still needed to some degree but their lines of delineation are no longer clear.
Traditional teams and silo’s now need a new breed of architect, the “Converged Architect” – one who understands the capabilities and limitation of the entire stack, holistically. Where to draw such a beast from is a challenge being played out daily in enterprises worldwide. All areas, network, compute, storage have needed skills and disciplines of merit, yet where to establish the foundations of your converged architect and what the nature of your support organisation is has yet to be fully defined for many – more of this another time…
Even before the changes demanded by the converged infrastructure are fully formed let alone established, we are seeing 2014 witness the rise and rise of the next iteration of convergence – the hyperconverged. With everything managed through the hypervisor, there is no physical SAN, or storage array; the networking is all virtualised, and there is only one piece of hardware. Procurement has suddenly become very easy. The interoperability isn’t just managed by someone else – it isn’t even needed – the pain resulting from the infamous interoperability matrix has become a thing of the past! Using a software-defined mentality, hyperconverged solutions, like those from Simplivity and Nutanix, appear to solve the lingering problems not fully answered by the converged solutions.
However, it isn’t quite sunshine and roses all the way. One vendor means lock-in. Lock-in means limited bargaining power. Also, implementing a hyperconverged solution means creating another silo in your data centre – haven’t we spent the last decade trying to stop that?
There is no doubt that there are significant benefits to deploying hyperconverged infrastructures. Procurement is simplified, scaling is more straightforward, plus support and maintenance is easier. The data centre also becomes a less complicated place, as physical routers, switches, SANs, storage arrays, and perhaps tape libraries, may all get replaced by banks of x86 servers. These are still emerging technologies, however, and are yet to be tested in the most demanding of environments where converged infrastructures (like those from VCE) and traditional architectures continue to excel.
So, clearly there is merit in considering converged and hyperconverged solutions. Not all of the marketing is hype. There is a fair degree of reality here, yet it does not represent a panacea and the complexities of deployment to production (rather than deployment to presence, i.e. the tin on the floor being used), still remain. For this reason the savvy CIO, who may well be starting the journey, will also be thinking of how these steps will address the wider business needs and allow them to maintain “today” whilst embracing the inevitable “tomorrow” – In the next blog, I’ll be taking a look at how we can achieve precisely that.
David Boyd, Director Accordant Solutions Ltd David.Boyd@AccordantSolutions.co.uk Mobile: +44 (0) 755 737 5993