In his recent blog post, "Don't dismiss cloud computing hype; creative fog is what makes cloud work", Kevin Fogarty of ITWorld asserts that cloud hype is actually beneficial since it stimulates innovation and adoption by driving providers to stretch capabilities and customers to try new things (a short paraphrase.)   While this would be completely true in an academic environment in which some new theory or knowledge domain was being debated and developed, there are negative consequences in the real world for both providers and customers of cloud.

Being on the frontlines of delivering cloud as a managed cloud service provider, I feel compelled to take the customers' side in this.  Sure, we as a provider have benefited from the hype, because it has brought us customers who thought we were the El Dorado of IT services, a land in which your every IT wish would come true (or at least the wishes made possible by all the hype.)  However he makes the implicit point in his article that the hype is good because potential cloud customers are too well-considered and conservative to fall for it, so it really benefits the cloud ecosystem by fostering innovation.   My experience is that such customers are few and far between - even IT management in larger enterprises often is squeezed and troubled enough by their current situation that they will grasp at solutions that might seem like hype to those more well-considered, and cloud buyers at smaller enterprises and especially entrepreneurs often think the hype is all completely real.  This puts us - as an infrastructure and platform cloud provider - in the position of having to compete with imaginary products like cloud services that are up 100% of the time.  Sure, it drives our innovation to achieve that service level, but what do we say to someone who thinks it's currently possible for any random cloud deployment?   In our case, we choose to educate as part of our sales process, even if the education process disappoints the customer or we find that they are best served by going elsewhere.

So it makes me wonder what happens to cloud customers who believed the hype and chose a provider that didn't bother to tell them that it was hype.  The results can't be pretty, as we saw a few months ago when Amazon had a data center failure that only surprised those who believed the hype, rather than reading Amazon's actual service level agreement.  So there lies the danger of the hype: it serves both the clients and the providers of cloud by building a shared delusion that allows them to achieve their business aims - until that delusion is pierced by reality.
This is very much the same situation as the '90s in which large enterprise suites were sold (and still are) with the hype that they will actually make the client company successful, instead of the all-too-common result of entrapping it in an endless miasma of deployment headaches.  There was substance behind enterprise suites, and they didn't fade, but they did disappoint large numbers of buyers who are now being experimented on by SaaS companies instead, though this time around the challenges are different, resulting more from integration issues than configuration issues.

As the holder of my customer's IT responsibilities, we take them very seriously and are looking - much like the much-maligned IT departments Kevin compares cloud to - to reduce their risks and eliminate the effects of the ecosystem of experimentation that he is praising.   Analysts are always talking about the watershed that will get large enterprises and those that think like them to move important parts of their IT to the cloud: perhaps it is realizing that cloud vendors aren't going to experiment on them with hyped features that will help them cross the chasm.

I just spent 3 days at the HostingCon conference in San Diego.   For a cloud service provider, it was pretty eye-opening to for the first time, see the entire ecosystem for hosters as opposed to the pure enterprise-cloud world that we normally work with.   What is amazing is that the bulk of the revenue from Hosting/Bulk Hosting/Shared Hosting still comes from individual servers running unvirtualized CPanel/Plesk or other control panels.   Many of the booths at the show offered add-ins or enhancements to CPanel to allow it to deliver more robust services, including better security, reporting, and load management.   There was even one vendor who offered a clustered CPanel setup promising high performance.  However, the dominance of control panels is being challenged by companies like OnApp, which offers an environment of easy virtual machine deployment with a control-panel like interface and hoster-friendly setup.

There were also a number of vendors showing SaaS services that complement traditional hosting.  Cloud Flare and a few other security-as-a-service offerings that replace hardware firewalls (including web application firewalls) seem to be showing promise at reducing the cost of security and DDOS mitigation.  Cloud Flare offers caching for javascript and images on DNS-optimized points of presence, essentially making it a CDN as well.   There were also some monitoring and billing vendors and third-party PCI certification companies, which reflected the increasing sophistication of both hosting companies and their customers.

A few hardware vendors came to show off their latest wares.   SuperMicro and Dell now offer mini-blades of up to 12 1-socket servers plus disk in a single 2U chassis.   DataOn showed a drawer-style JBOD system similar to Sun's old "thumper" system, with very high density.  AmpliData has an object storage system with high performance and lower power density.

What was missing from the show was any kind of "real" enterprise-style cloud product, offering virtual private datacenters and management capabilities based on hierarchical customer entities.   It appears that both hosters and their customers are still happy with the traditional hosting model, leaving cloud computing to upmarket and vertical-market ecosystems.

This meant that I didn't really have a compatible language to describe to people I met what it is that ENKI does.    I finally came upon the idea to use old (hosting-age) terminology to describe the new generation of cloud services that ENKI offers:

dedicated server -> virtual dedicated server

colocation -> virtual colocation  (a customer-managed virtual private datacenter)

IT services -> virtual IT services (provider managed production IT in the cloud)

With this Rosetta Stone based on the concept of virtualization, I was able to get some good conversations going with hosters about the similarities and differences in our products and approaches, and how we could serve our customers better.   In particular, many cloud customers start with hosting and/or VPS services (especially as they prototype their websites), necessitating a strong cross-pollination of hosting and cloud vendors and services in order to better serve the market and avoid the customer disasters that Amazon experienced a couple months ago.

In addition, I saw that our private cloud deployment service, PrimaSys, based on our Computing Utility, is a good match for hosting providers looking to move into the cloud world, since it offers turnkey VPS provisioning, while also allowing whole applications to be deployed with a single click.