I recently had a conversation with a stealth cloud startup called ComputeNext which got me fired up because they are doing something that I have been advocating here in this space, emergence of an open federated cloud ecosystem. Well, they are not building a federated ecosystem per se but they have gone one step further to make search and discovery of compute resources in a federated ecosystem much easier. In this post, I will highlight the problem they are trying to solve and their solution. However, keep in mind that they are in stealth mode and I could only get some high level information about their technology.
The Problem
Regular readers of this blog know that I am an unabashed evangelist of an open federated cloud ecosystem. Conventional wisdom predicts a consolidation of players in the infrastructure space because not many companies that can scale to meet the world’s needs with margins in infrastructure being very low. But I have given both philosophical and economic reasons about why such a consolidation will not happen and have also pointed out to specific instances that will require a more federated ecosystem in place. With the success of cloud platforms like Eucalyptus, Cloud.com, Cloupia, etc. and the anticipated success of OpenStack, a federated cloud ecosystem is fast becoming a reality.
However, when we have such federated resources across different geographies, different service levels, different compliance levels, etc., resource discovery becomes a big problem. We don’t have a reliable way to search and find compute resources in a world with federated cloud ecosystems similar to how Google helped us find the right information from the distributed web of documents in the internet. SpotCloud and ScaleUp Technologies have come up with interesting ways to enable a marketplace for resources in federated cloud ecosystems. But I would like to see a Google like holistic approach to resource discovery in the federated cloud which will let me consume compute resources by slicing and dicing them in the way I want based on either geography or service levels or compliance needs, etc..
ComputeNext Solution
ComputeNext (remember, they are still in a stealth mode and don’t even have their website ready yet) is approaching this problem in a holistic and transparent way. They aggregate resources not only from smaller cloud providers but also other traditional datacenters by on-boarding them into the federated cloud model by using one or more of the existing cloud platforms. For instance, these traditional datacenters whose business is slowly getting disrupted by the onslaught of public clouds can tap one of the cloud platforms like Eucalyptus, OpenStack, etc. and be part of the federation aggregated by ComputeNext’s technology.
ComputeNext can aggregate resources from heterogenous environments and from datacenters of all sizes and tiers. These resources are then offered to “consumers” through ComputeNext console which lets them search, discover and rent compute resources based on their needs. A virtual appliance, called ComputeNext Federation Server, sits between the console and the datacenters speaking the language of the cloud platforms used by these datacenters and providers to aggregate, monitor and manage resources on both compute and storage side. The user (they may be an enterprise or even another cloud provider needing additional resources to meet the demand) can use the console to search for their needs and easily find the resources that exactly match their requirements.
We always underestimated the importance of search and discovery until Google showed us how it is an important problem in any distributed ecosystem. As more and more countries join the cloud bandwagon with their own public clouds, the emergence of regional cloud providers are a reality. Also, with a potential to tap into mobile and cloud technologies to reshape Africa, we will soon be seeing datacenters in that continent serving users from different countries in Africa. Microsofts and Amazons of the world have no business interest to set up shops datacenters in Africa and we will only be seeing regional players setting up cloud infrastructure using technologies like OpenStack. With all this becoming a reality faster than we expected, the new conventional wisdom should be about federation and not consolidation. ComputeNext is positioning themselves to be right in the middle of this federation.
My POV
Let me add few of my thoughts about ComputeNext here
- I got extremely excited hearing from this stealth startup because their plans fit well with the ideas I am promoting here and elsewhere about open federated cloud ecosystem
- There is definitely some overlap between ComputeNext and the recently launched SpotCloud. However, ComputeNext has a slightly different goal and execution. The first difference between ComputeNext and SpotCloud is that ComputeNext don’t have their own cloud platform whereas SpotCloud has Enomaly platform. Instead, ComputeNext’s technology works with cloud platforms from other vendors. The second major difference I see between ComputeNext and SpotCloud is about transparency (a point which I highlighted in my post on SpotCloud). ComputeNext wants to take a more transparent approach because they feel it is the only way they can get the enterprises to trust public clouds. Users can select their resources based on geography, SLAs, compliance requirements, tiers, etc.. SpotCloud with their opaque approach can easily target smaller providers but ComputeNext will have to focus only on datacenters who will be willing to share their information with potential buyers
- Some of the ideas they discussed with me on the potential to take their service to the next level really revved me up about what they can do in the future. The kind of automation they can potentially bring to a federated cloud ecosystem can literally change the way we are seeing cloud computing today
- Having said that, they have their task cut out before they could become the Google of compute resources. They need to convince vendors like Eucalyptus, Cloud.com, VMware, Cloupia, OpenStack, etc. to interoperate with their technology. More importantly, they need to convince traditional datacenters to morph into a cloud provider and on-board them along with other regional cloud providers to be part of their aggregation. It is no easy task. When I pressed them hard on the difficulty, they acknowledged that and told me that these datacenters will be more forthcoming once they see the value in their offering.
It will be interesting to see the evolution of not just ComputeNext but also the federated cloud ecosystem with players from all around the world.
Your comment “they have their task cut out before they could become the Google of compute resources” may be an incorrect analogy.
A better analogy is a Visa/MC network or a stock exchange. Given data centers need to join and deploy proprietary software from ComputeNext it resembles more of a closed market place than the www upon which a Google search model flourishes. For example, unless Spot Cloud makes its resources available, which I would presume requires deploying ComputeNext technology, it would not be searchable nor discoverable.
I see Computenext as a Confederation, not federation. It is an elastic membership for data centers that can turn a number of Confederation members into a federation.
I agree ComputeNext is confederation but I used the term federation to denote the ecosystem and not ComputeNext. re: wanting to become Google of compute resources, yes it is not a perfect analogy but what I am trying to give is a feel for their service. I didn’t use Stock Exchange analogy because of the emphasis on search & discovery than the marketplace angle of it.
Also, given a chance, they would love to aggregate all of world’s resources. I wouldn’t take away the Google analogy because someone is not allowing them to index because I could still do it to Google using robot.txt on my side. Anyhow my use of Google is to highlight the importance of search and discovery on their service.