VMware has made recent changes to EVO:RAIL licensing that promise to draw in users. This is a welcome -- and frequently demanded -- change, but it may not be enough for VMware to dominate the hyper-converged market.
One of the sticking points about EVO:RAIL has been that the licensing must be included in the sale. Until recently, this meant that buying an EVO:RAIL appliance meant buying vSphere licenses -- licenses a company may or may not actually need.
After prolonged hammering on this point by hyper-converged infrastructure rival Nutanix, VMware has softened its stance somewhat. If you have obtained licenses via Enterprise Licensing Agreements (ELAs), OEM partners, distribution, or other resale channels, then you can now apply those licenses to your EVO:RAIL appliance instead of having to buy new ones.
This does not mean you buy EVO:RAIL without licenses. VMware still insists on ensuring EVO:RAIL is licensed, but this adjustment has gone a long way to make EVO:RAIL palatable to enterprises. Nobody except SMBs are actually forced to pay list price for software; the cost per node for EVO:RAIL was significantly higher than enterprises were used to paying.
What's more, EVO:RAIL is frequently brought in during refreshes. Old hardware goes out, new hardware comes in, and the licenses move from one to the other. Being forced to buy additional licenses could leave enterprises with a surplus, and that makes both the procurement and IT departments grumpy.
The softening of the licensing stance will help VMware going forward, but it has already earned some ill will. Nutanix managed to capitalize on the issue effectively. Now that VMware has dropped the requirement, it is hitting back by promoting examples where EVO:RAIL is cheaper than Nutanix.
Both parties claim performance, ease of use and simplicity leads over one another and are also beset by innumerable other hyper-converged vendors all vying for mind -- and market -- share. VMware, with its newfound licensing changes, huge sales channel and massively popular brand name should be dominating the hyper-convergence space handily. Currently, VMware has not managed to rise to the top.
It's about more than price and performance
The traditional wisdom in IT says that if you have the fastest and lowest priced widget, you win. Issues like ease of administration, ease of setup and ease of maintenance are traditionally ignored because systems administrators are viewed as disposable. Executives can simply whip the workforce until they comply.
This attitude is changing, and hyper-convergence is one of the leading technologies to take advantage of this shift.
Ease of use matters. Everyone from end users to executives has heard of public cloud computing and many have seen exactly how easy it is to spin up IT solutions in such an environment. This realization is rippling through many data centers.
There is less and less tolerance for projects that span several months to implement IT projects. Installing, configuring and testing applications takes time. There isn't much any IT department can currently do about that. This places the pressure on infrastructure teams to cut the time between a request and running the environment down to nothing.
Here, Nutanix -- and many other hyper-converged players -- have several advantages.
Other hyper-converged companies integrate some or all of their management interfaces into vSphere via plug-ins, and into ESXi via kernel modules. This is gated somewhat by the speed at which VMware chooses to certify both for use, but the end result is that in practice the various hyper-converged solutions work seamlessly from within vSphere.
VMware offers analytics, monitoring and so forth as a separate product: vRealize Operations (vROps). Other hyper-convergence vendors offer that functionality as part of the basic product. Some, like Nutanix, have made the advanced state of their monitoring and analytics a key selling feature.
Another ease of use element comes with configuration and flexibility in cluster diversity. EVO:RAIL is quite restrictive; CPU, RAM and storage amounts are fixed. There is no ability to vary which CPUs are used, how much RAM there is pre node or what the flash/disk mix is. Other hyper-converged vendors offer this ability.
Diverse node cluster support is also missing with EVO:RAIL -- you cannot mix and match nodes of different configurations within a cluster. Indeed, you can only buy EVO:RAIL in four-node blocks so expansion of a cluster is done in big jumps.
To use Nutanix as an example of how other vendors approach this, you technically only need three nodes of any configuration to make a Nutanix cluster. If you needed only a little CPU and memory but need a ton of storage you could theoretically run one of the company's 1065 nodes or two of its 6035 nodes. In practice, you'd start out with three identical nodes then add different nodes of different configurations on an as-needed basis as the cluster grows.
There are plenty of other examples of how the various hyper-converged vendors are working to pack as much functionality and flexibility into their offerings as possible, but hopefully the point has been made. Flexibility and functionality are the new battlegrounds, and they are occurring at higher layers than the hypervisor itself. If VMware is going to win the hyper-converged wars, it needs to be prepared to change.
Uncover the differences between VMware’s EVO:RAIL partners
Things you should know about VMware EVO:RAIL
How EVO:RAIL compares to other hyper-converged options