The benefits of hyper-converged infrastructure are largely based on the simplicity of the systems when compared with other, more traditional approaches.
Instead of purchasing hardware and software from a variety of different vendors and them attempting to get them all working together, a pre-configured system is purchased from a single vendor.
The system scales horizontally, so if additional storage or compute capacity is required, you simply buy another node and add it to the existing cluster, without an outage. Software upgrades are also done online, usually by downloading the update from the vendor over the internet, like a simple app.
Hyper-converged infrastructure is another way in which traditional data centre kit is being transformed by the consumerisation of IT as people – used to the simplicity and ease of use of smartphones – come to expect the same experience from all of their IT.
This ease of use is particularly appealing to the small- and mid-size IT shops that don’t have scores of dedicated specialists in each part of the IT stack. Instead, teams of generalists are expected to handle compute, storage, networking, databases, and applications. All at once. If these systems are complex and fragile, they take up a lot of administrator time. Time that could be better spent helping their organisations provide a better customer experience or increasing revenues.
But with the simplicity of hyper-converged systems that require little human intervention, do customers really need the same size IT team? Will customers reduce staff in their IT departments, or will these existing employees be redeployed to work on other things?
CRN asked resellers, distributors, and customers for what they’re seeing out there in the real world.
What is hyper-converged infrastructure?
While there is no clear ‘official’ definition of hyper-converged infrastructure, a working definition says this technology comprises commodity x86 servers with storage and compute in the same chassis, configured as scale-out clusters, with a hypervisor providing hardware abstraction. Virtual machines run on hyper-converged infrastructure, allowing them to move around the cluster sharing the available resources. Adding disk or CPU capacity is done by adding nodes to the cluster.
The primary advantage of hyper-converged infrastructure is simplicity. The nodes themselves are pre-configured units, purchased from a single vendor who supplies the hardware, software and support for the system as a whole. Installing them is generally as simple as racking the systems, plugging them in and turning them on. There are very few configuration options. Adding capacity is as simple as plugging in another node.
The trade-off for this simplicity is a lack of nerd knobs for customers to twiddle in their attempts to eke out every last skerrick of performance. Hyper-converged vendors are betting that a lot of organisations don’t need highly tuned systems, and instead just need some computer to get their work done. The sales growth of the major hyper-converged vendors suggests that they’re probably right.
Big business
Hyper-converged is a hot topic right now, with vendors boasting of billion-dollar valuations and hundreds of millions in investment poured into their coffers. Nutanix took $140 million in investment in August 2014 and reports having more than 1,200 customers, while Simplivity slurped up $175 million in March 2015 and has just over 550 customers. The SMB-focused Scale Computing added another $18 million in investment in July 2015 and boasts of more than 1,000 customers.
IT behemoth VMware entered the market with its EVO:RAIL reference architecture at VMworld 2014, along with a host of vendor partners ready to sell gear made to the EVO:RAIL specification, including Dell, Supermicro and Fujitsu.
VMware parent EMC released their take on the EVO:RAIL platform, called VSPEX BLUE, in February 2015, however industry gossip suggests the EVO:RAIL platform isn’t selling as well as originally hoped. HP recently dropped its support for the program, less than a year after joining it in October 2014.
Other vendors jumping onto the hyper-converged bandwagon in recent months include Atlantis, which has started offering its own OEM branded hardware to go with their software, and Maxta, which bills its software as enabling hyper-converged systems by installing it on hardware you purchase yourself (or through a reseller). Then we have a smattering of other, minor players including Nimboxx and Pivot3.
Next: What's happening in the local scene
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
