Blog Post

Can a New IaaS Cloud Really Compete?

New cloud provider NephoScale announced its presence among IaaS providers earlier this week, touting itself as “an advanced cloud service for serious programmers.” However, as I explain in my weekly update at GigaOM Pro (subscription required), I’m afraid its message might fall upon deaf ears, as there’s little evidence the world is clamoring for another IaaS cloud.

For new entrants into the IaaS space, the biggest obstacle to getting attention is the sheer number of players operating within the United States. A nowhere near exhaustive list includes Amazon Web Services, Rackspace, Microsoft, GoGrid, Skytap, Linode, Slicehost, OpSource, Terremark, Bluelock, Savvis and AT&T. And many of these companies are already well respected, having developed strong reputations as cloud providers, MSPs or both. Several of these vendors — AWS, Rackspace and GoGrid, especially — have broad feature sets (whether developed internally or achieved through partnerships) enabling everything from hybrid dedicated-virtual clouds to running Hadoop applications.

APIs are an interesting point of comparison, too. In an interview with HPC in the Cloud, NephoScale Founder and CTO Telemachus Luu explained its CloudScript product in this way: “Leveraging NephoScale’s CloudScript, a user can deploy an entire datacenter configuration through a single API submission (JSON post).” Although this might be a technically unique API, part of luring developers is having the best community and the broadest support. Some already consider Amazon’s API a de facto standard, and the Rackspace-led OpenStack project is bringing in developers by the thousands. It will take a very impressive API and overall service to overtake this type of momentum.

Finally, there’s the question of whether IaaS clouds will even be the primary stomping grounds of skilled developers going forward. Conventional wisdom in the cloud community is that PaaS, which lets developers write applications without worrying about infrastructure-level concerns like servers and storage, is the future of cloud development. From newcomers like PHP Fog to established vendors like Red Hat, PaaS is attracting much attention and much investment, and the reason is its promises of disrupting application development.

I’m not arguing that Nephoscale — or any new IaaS provider trying to compete on a large scale — can’t succeed, just that it will be very difficult to do so. With on-demand VMs being all but a commodity at this point (see Enomaly’s SpotCloud, for an example of this proposition), features are what set cloud providers apart — and the existing pool has had years to build them and have plenty of money to build or buy more. New IaaS providers need to grow up fast or go PaaS.

Read the entire post here.

Image courtesy of Flickr user shawnzrossi.

Related content from GigaOM Pro (sub req’d):

5 Responses to “Can a New IaaS Cloud Really Compete?”

  1. We are in the boom phase of cloud computing. There is room for many entrants and tons of differentiation still. Here are three areas that are so early it’s hard to find public clouds that do this:
    – Network and network services.
    – Storage with differentiated storage levels and latency.
    – PaaS that bridges the old apps to the new infrastructure.

    There’s more and the notion that cloud is what we in the valley use for our startups is applicable to all the world leaves a lot on money on the table.

    There will come a moment of reckoning for this capacity. But not the next two years.

  2. “Another web hosting company? How can it compete?” And yet, every few years, a new webhost will show up and do pretty good business for itself. In an industry far more stable and mature than the cloud offerings out there.

    ANY new IaaS cloud service has a tougher row to hoe, sure – and yet, the field has already been blowed for them as well. It’s easier and easier to get a new service off the ground, and with new projects like OpenStack and older projects like OpenNebula and CloudStack, there’s tons less work to do. Users mostly know what they’re getting as well. Yup – new offerings need something to compete with. Can be price, or service, or features, or marketing… but telling the world that the market’s already locked up is like saying the PC market is owned by IBM in 1982.

  3. To the point Wes made, it is silly to rule out a new entrant into the IaaS layer such as but not limited to NephoScale. This is also an irrelevant question posed:

    > Finally, there’s the question of whether IaaS clouds will even be the
    > primary stomping grounds of skilled developers going forward.

    Whatever. Skilled developers should not also possess some IaaS admin skills? We’re talking about a layered stack:

    Besides, no one in their right mind would every allow themselves to be held hostage with vendor-lock in at any of these layers. Or maybe the apps developers don’t care since all they want the responsibility of doing is building the apps for their masters who will run the businesses which will provide them with their salaries (whereas their daddies will be the ones who have to look after the lower levels of the stack).

  4. I suppose if you’re just copying Amazon you will find the IaaS market saturated. But I also suspect there is a silent majority of customers who want to run their existing software on the cloud but can’t because of its artificial limitations compared to real servers. Using JSON is hardly a differentiator, but that doesn’t mean no differentiation is possible.