This month, SimpliVity Corporation (www.simplivity.com) announced a Series D round of financing, raising an additional $175m for a total valuation of $1bn. To date, the company has raised $276 million in four rounds (press release). The money is to be used to fund company expansion, including a global roadshow event at which I will be presenting on behalf of Langton Blue (disclosure: SimpliVity will be a client of Langton Blue in May). It is clear that hyper-convergence has struck a chord with customers, no doubt due to the sheer simplicity of the solutions.
Nutanix, arguably the leader in the hyper-converged space (if not by any other measure than being a first mover) has raised $312 million to date and is pushing hard on expansion with a heavy emphasis on marketing. The company has recently announced their own certification programme (press release) and will hold their first conference in June this year (link). Both of these companies are marketing hard to gain mindshare with potential customers before the market becomes more diluted with entrants like VMware (EVO:RAIL), Scale Computing and a host of software only offerings (e.g. Maxta, Springpath).
Is hyper-converged the first step on the road to web-scale computing? Making that assumption may be a step too far, but hyper-converged is definitely changing the deployment model for technology. Even with virtualisation at the level we see it today (perhaps 70-80%), there is still too much focus on the concept of a virtual machine as the unit of compute, harking back to a rigid view of physical hardware. Hyper-convergence starts the move to a more abstracted way of working, reducing hardware to the level of the server, without a dependence (or at least less dependence) on the physical aspects of that server in the deployment of the application.
Where we go next will be interesting to see:
- Hyper-converged solutions need to scale massively to become web-scale platforms. This means becoming (physically) leaner than today’s offerings and harnessing new hardware solutions like OCP (Open Compute Project). The same applies to the software components.
- We need to lose our dependence on VMs. Workloads need to move to containers and applications need to run across containers on physically separate platforms. The O/S layer needs to become as thin as possible (like CoreOS, RancherOS) while still retaining key features around security and management.
Both of these visions require huge changes at the application and data management layers. This is where I see the next set of challenges will lie. Hyper-convergence is a stepping stone to the next level of application optimisation. All we need now is a good name for this new paradigm.
Note: Langton Blue will be publishing an independently authored white paper covering the Hyper-Convergence market in April 2015.