Just read an interesting article on Forbes.com by Dan Woods entitled “Parsing the Cloud“. Dan makes a similar argument to our own Ian Rae, suggesting that specialized clouds will be required to meet the privacy, regulatory, geographic latency and application architecture demands of cloud consumers.

This begs the question, who will build all these specialized clouds? Are there incumbents who simply need to evolve, or will we see the birth of dozens or hundreds of new cloud providers?

You only have to Google “HIPAA Hosting”, “Sarbanes Hosting”, or “Swiss Hosting” to discover that there are a plethora of managed service providers (MSPs) offering hosting solutions that comply with specific regulatory requirements.

Specialization versus economies of scale

Clearly there are many hosters in almost every region of the planet, capable of delivering fast service to local users. And there are many hosters that specialize in specific applications or application architectures. These hosters have an existing relationship with application owners that need their specific brand of hosting, putting them in a good position to evolve to meet the desire for more flexible, scalable hosting options.

What these incumbent hosters are unlikely to be able to provide, however, is the low cost of massive multinational cloud providers like Google, Amazon and Dell that in many cases build their own hardware to keep costs as low as possible. On the other hand, these big cloud providers lack the specific vertical knowledge and customer relationships required to effectively deliver specialized services.  With this in mind, I see a few ways this vertical cloud market can develop:

Option one: Vertical hosters make their own clouds

Vertical hosters can adopt cloud computing platforms like VMWare vCloud, Citrix C3, QLayer, 3tera, Elastra, Enomalism, Vertebra and ReasonablySmart, to transform their own data center into a cloud. This scenario would be great for the aforementioned vendors and would keep a lot of small hosting companies in business. I think some of this will happen, however it will only survive long-term in small niche applications that have architectural and regulatory constraints that can not be satisfied as a layer on top of a commercial cloud no matter how distributed and controllable it is.

Option two: Specialized clouds inside generic clouds

Domain experts, which could include the existing vertical hosters, could layer on top of large clouds like EC2 to create vertical offerings. This solution sounds appealing, but would not satisfy many of the requirements that are driving the need for vertical clouds. Large cloud platforms would have to mature a great deal to give vertical cloud providers the ability to self-manage elements like privacy and data locality. Essentially, EC2 would have to become a Ning-like platform for white-labelled private clouds.

This is a model familiar to incumbent hosters who have long used provisioning platforms from companies like SWSoft/Parallels that support white-labelled hosting businesses, however these shared hosting platforms are far simpler than what would be required for a global cloud. Near-term, I believe there will be many vertical layers on top of commercial clouds.

Some simple offerings such as Heroku on EC2 for Ruby development are already available. However, I think it will be a long time before big cloud platforms will be Ning-like enough to offer vertical cloud providers the control they will need to satisfied their constituents. If you run a regional hosting company, find a speciality ASAP that can’t be easily deployed on large clouds due to privacy, regulatory or architectural constraints. Ruby development and deployment probably isn’t it.

Option three: Big cloud players roll up the landscape

Finally, large cloud providers may acquire existing vertical hosters that specialize in the larger verticals such as HIPAA, Sarbanes, and Swiss Banking and in large geographies with nationalist policies. They will blend their sophisticated cloud management platforms with the domain knowledge, reputation and customer relationships of the vertical hoster to dominate the specific target segment in both price and capability. It is unlikely that market pioneering companies like Amazon and Google will lead this charge however. Rather, companies like IBM and HP/EDS will likely blaze this trail as they have extensive experience with most geographies and verticals already.

It is possible that these companies may try to build these vertical hosting solutions on their cloud platforms organically too, they do have data center presence in many geographies already. But history would suggest they will eventually engage in an acquisition frenzy as the land-grab (or should I say cloud-grab) accelerates. They’d get to expand their market in each vertical while denying the cloud pioneers access to domain specific skills and customers  (Eventually Amazon and Google will realize that the margins are in vertical clouds, not in sucking VC dollars out of W2.0 companies with a 1 in 10 survival rate).

So what’s going to happen?

It’s always hard to predict the future. But here’s how I think things will play out in the coming years.

• Short term there will be a spat of vertical hosters tooling up with cloud platforms, or moving their operations onto a larger commercial cloud if their business allows. During this period, MSP’s with a speciality not easily deployed on a generic commercial cloud will do alright if they can offer cloud-like services.
• Medium-term will be the entry of the IBM’s and HP’s, building and acquiring strong vertical and regional MSPs, but likely only serving the highest end of the market. MSP’s with a strong speciality may find themselves the belle at the ball during this period.
• Long term, though, (5+ years), I think it will be the battle of the globally dispersed white-label clouds with highly controllable Ning-like platforms that will enable most any special needs to be deployed on their infrastructure. That’s when I wouldn’t still want to own a regional MSP or vertical hosting company.

Very interested in other views. I look forward to your comments.