Nick Carr, in his book The Big Switch, used the electricity analogy to explain the nature of Cloud Computing. Initially, this comparison helped people get enthusiastic about cloud computing by connecting the idea with that of electric generation. However, I think this concept has outlived its utility and we need to go beyond this simplistic model. Recently, I had a twitter discussion with Randy Bias of CloudScaling on the topic. He didn’t like the idea of using the term outsourcing while describing Cloud Computing. I thought there is a need for this term because, unlike electricity, we cede lot of control to the cloud provider. However, Randy disagreed and emphasized on the comparison to the electricity model.
@krishnan Whoa. Disagree. You give up control of power generation. You become a self-service consumer. *exactly* like cloud.
Already, James Urquhart has written an eloquent post on the topic. But, at the risk of appearing to be repetitive, I want to address the topic here because the 140 character limit on Twitter is too restrictive for such discussions.
Let us do some comparison between cloud computing and electricity model and see where these two resemble each other.
- Electrical power generation and delivery through electrical grid are similar to the cloud provider taking care of compute power and delivering through internet.
- Pay as you go pricing model.
- Enormous cost savings that accompanies large scale centralized power/compute generation and delivery.
In a way, the comparison ends here. As James clearly highlighted in his post, when you put the all important data into the mix, everything changes. There is no parallel to compute data in the electricity model. The various issues surrounding the data clearly limits the comparison to the electricity model. Even though we let go of the control to manage computing infrastructure much like how we let go the control of power generation, the presence of data inserts the additional outsourcing component to the definition of cloud computing. This is due to various risk factors, security and regulatory issues that come up when we throw the data into the mix. There is no analog in the electricity industry that matches the risk factors introduced with the transfer of control of data in the case of cloud computing.
However, this doesn’t mean the electricity comparison is invalid. On the contrary, it is still the most attractive idea in cloud computing. It also doesn’t mean that we should keep away from cloud computing. Rather, we should embrace cloud computing for the electricity-like benefits but we should consider various issues surrounding the data while planning for the cloud adoption. It is important for us to realize that there is an outsourcing component involved here and we need to use due diligence during the planning stages like any other outsourcing process. The electricity model is a good starting point for understanding the advantages of cloud computing but we need to go far beyond this model to implement it. Feel free to jump in with your take on this topic.
Nice post. It might be that some are railing against the “oversimplification” between electrical utilities and cloud providers? While I don’t like oversimplifications either, I do agree that it’s a good starting point in understanding the possibilities – electrical utilities are infrastructure – pure and simple. Cloud is infrastructure as well. Most of the active discussions/topics are from the perspective of people supplying to or building “clouds”, so the perspective is one of the “transformer manufacturers” or the “wire suppliers” – a bit skewed and certainly removed from where the ultimate value will be. For now, though, the bulk of the value around clouds is in building, connecting to, securing, and enabling the deploying to/on clouds. In this landscape too much weight is given to developers, but over time, when an “average Josephine/Joe” can solve a problem and not have to think about the technology behind it, THAT’s when cloud will be at the value end of the scale…
My two cents; no change required!
Rich
My point on twitter tonight to SamJ was pretty basic around this. If #cloudcomputing is a lot like electricity we should see an increasing level of standardization around compute, storage and networking options over time. Instead we have seen AWS release more exotic instance sized (64gig RAM) etc designed to work best for specific applications types and use cases. This is a simple, but deadly difference that ends the analogy dead in its tracks as far as I can tell. The only similarity remains part of the output being delivered over what resembles a wire.