Cloud service models of today?

The last years we have all read, and talked about cloud computing. That everything will be executed in the cloud … To some extent that is probably true, but for an enterprise there will be systems that there are no business case in moving into the cloud, as well a there are systems that definitely needs to be executed from a cloud platform.

Many different types of cloud will exist, and I argue that a company benefit from selecting a mix of these.

The first, and most simple version of cloud is the hardware cloud (IaaS or Infrastructure as a Service), the characteristics of this cloud is that you buy the amount of computing / hardware that you need for the moment, and can release capacity that is not needed with very short notice. In a perfect world the scaling up, and down would be automatically and in real time, and depending on what service level agreement (SLA) the buyer have with the cloud vendor.

e.g. if performing for example computing intensive supply chain planning calculations the cloud would understand that to keep the SLAs it need to scale up the allowed performance. This sometimes also is called “cloud burst”. One big difference in the IaaS cloud is how you pay for your services, the most common is that you pay for allocated resources, but some cloud providers actually invoice per used resource unit. If the cloud bursting shall work it is very important that the invoicing mechanism is per usage and not per reserved capacity.

The next common abstraction level in the service model dimension is the PaaS or platform as a service, here we have abstracted away the hardware, and get access to a computing platform like a database, or a web server or a execution runtime engine. When reaching this abstraction level it can be interesting just to quickly reflect on the differences of IaaS and PaaS from a cloud owner perspective. The IaaS will most likely be owned by the hardware vendor, that owns and manufactures the computing platforms, or the storage. The margin that these vendors can play with to get the correct price is directly dependent on the hardware. In the second case the PaaS we will on top of some hardware add a layer of executional functionality like a database or a runtime execution engine on where some kind of software will execute.

When we look at the third abstraction layer of cloud in the same dimension we will come to the SaaS or software as a service level and this is where the service offered is access to a piece of software. Examples of this can be email, CRM, and even ERP systems. Now we are looking at a cloud that is handled by a software company. The margin that now is regulating the sell price of the services offered by the cloud vendor comes from the licenses that a using company have to pay. The hardware is still needed, as well as the platform to execute the software upon. We now have three different service cloud models that spans from one extreme of hardware only to software only. I will now argue that this is not the end of this dimension. To come close to the end we need to add a fourth cloud type that I will call the industry solution cloud (Ind aaS, Industry solutions as a service)

The highest business value is received when a company selects to run a industry solution that is standardized and managed in a separate cloud, a Industry solution cloud. In this end of the service dimension the cloud provider will by definition own a solution that is a compilation of several software entities that together result in a higher value to the end customer than each piece of software on it own. To be able to run this successfully the added value need to be higher than the cost of software license, platform cost and hardware cost.

So far I have not seen any clear provider of industry cloud, but I am sure that they will soon turn up. When there are a white spot in the market it will be filled by someone very soon, but the reason that we don’t see anyone here yet might be that there are no mature offerings that have a good enough business case. It is quite easy to argue against the IndaaS, for example the most common argument that I have heard is that you will not get any differentiation to your company if you chose this route. I believe that the IT industry in mature enough to understand that a solution in a IndaaS will have to be a set of black-box components that have well defined interfaces and can be exchanged for some other black-box component that will interact with the rest of the components in the same way. One ERP system that is closing in, and can be argued to play in this area is the Microsoft Dynamics AX where you have industry solutions, and also a powerful workflow engine where you can exchange sertain functionality in the standard ERP system with your own unique solution that would give the enterprise that key advantage that they need over their competitors.

Sometimes I use the term “Vendor Cloud” to describe a type of cloud that is in between the SaaS and the IndaaS, example of this could be a ERP vendor that have a cloud where the customers can run the ERP suit of software. This definition limits the control to the vendors software, and hence it still is a license play when it comes to cost, but some very large software vendors own suits of software that makes it close a industry cloud, but I think the difference is in the fact that an ERP system in it self do not give a business value, the business value is received when all the parameters, and most external integration are in place.