In-Depth

Time for a Virtualization Marketplace?

To keep up with cloud providers, the answer is yes.

• By Trevor Pott
• 08/19/2016

One aspect of my day job is to advise startups on how to make it out there in the wild world of the virtualization industry. One of my most frequent points of analysis is whether or not what the startup has created is, in the medium term, a feature or a product. Increasingly, the lack of a marketplace is a locus of discussion.

When I'm talking about a marketplace in the context of virtualization, I'm talking about a marketplace for buying or subscribing to virtual appliances. Many public cloud providers have them; you can deploy all sorts of workloads on a public cloud with just a few clicks. Some of these are free to use, and some must be licenced.

A virtualization marketplace is, essentially, an app store for virtual workloads.

While it might seem obvious that this is a useful tool, it's surprising how much resistance there is to the concept. Marketplaces are a lot of work, and they can be a significant expense. The administrative effort far outweighs the technical requirements, and it isn't something a lot of startups are looking to commit to. Despite this, I fear that those who don't choose to built a marketplace will be lost to history.

Why a Marketplace?
The biggest reason I advocate the creation of a marketplace is the "public clouds have it" argument. If you happen to be a hyper-convergence vendor, hypervisor vendor or anything similar, then your biggest competition isn't other virtualization vendors; it's public cloud vendors.

On-premises solutions are fighting for their right to exist at all, and at a time when public cloud is going from strength to strength. The costs of the public cloud are plummeting, while on-premises vendors raise prises to compensate for lost clients.

With this as the backdrop, it should be pretty clear that on-premises vendors simply can't afford to fall behind the public cloud. The public cloud is already far easier to use than on-premises solutions. As the price of the public cloud approaches that of running one's own equipment, that ease-of-use factor will win the day.

The three biggest advantages the public cloud has from an ease-of-use standpoint:

1. Turnkey operation (which hyper-convergence has solved for on-premises solutions)
2. Self-service portals (there aren't a lot of good options on-premises yet, and none are affordable)
3. The marketplace

Marketplace Obstacles
The main roadblock to the creation and maintenance of a virtualization marketplace is the issue of standardization. Amazon's marketplace works because Amazon is huge. They can afford to maintain their marketplace on their own and to populate it with a number of things their own engineers have devised.

Third parties want to get their wares on Amazon's marketplace because of that size. In the mobile market, everyone writes for iPhone and Android; Windows phone has to beg developers to give them the time of day.

With Amazon, Microsoft and Google clouds offering a lot of potential customers for minimal effort, third-party application developers aren't interested in every hyper-convergence vendor lighting up their own incompatible marketplace. The solution to the marketplace problem in virtualization isn't three dozen Windows Phones.

Survival Skills
The short version of all this: if on-premises virtualization vendors are to survive, they need to get over their innate "not-invented-here" syndrome and work together to form a standard on-premises application marketplace. That marketplace needs to be as attractive to third-party application designers as Amazon's, because it covers all the on-premises installations.

Microsoft is ready to extend their Azure market into the datacenter via Azure Stack. Docker is ready to offer ease of use that traditional virtualization providers can't match. Will VMware and the various KVM-based solutions come together for mutual survival? Or die with hands around each others' throats?