Public versus private clouds: why one size does not fit all
There has been a lot of discussions these days about private and public cloud. More discussion has been generated because both Amazon.com and Salesforce.com have added a Virtual Private Network (VPN) option to their public cloud services. What does this mean in the context of how customers will move to cloud computing? It is clear from the research that I have been doing that the private cloud and the hybrid cloud are real and will be part of the computing landscape for a long time. The emergence of the virtual private cloud is an early indication that customers some customers want a better guarantee of their data. The combination of a public cloud with the privacy offered by a VPN is only going to grow over the coming year.
So, is a Virtual Private Cloud still a public cloud? I particularly found the blog published by Amazon’s CTO,Werner Vogel’s announcing the virtual private cloud fascinating. On one hand, the private virtual cloud announcement is a proclamation that customers want to be able to have secure access to services on the Amazon EC2 Cloud. On the other hand, he is quite clear that this there is no such thing as a private cloud. Clearly, it is in Amazon’s best interest for customers to focus on public clouds. Vogel states in his blog that “What is called private clouds have little of these benefits (he means characteristics of the cloud) and as such I don’t think of them as true clouds” The four characteristics of the cloud he points to include:
- eliminating costs – lowering both capital expenses and operating costs
- elasticity – avoiding complex procurement cycles and improving time to market
- and removing undifferentiated heavy lifting by off loading data center operations
While I agree that there are many situations where this is an ideal approach for many businesses, I don’t think the situation is black and white. There are indeed shades of gray. In my view, a private cloud has to be architected to be different than a traditional data center. But like a traditional data center, it is protected by a firewall and sophisticated security. A private cloud will almost always be combined with some public cloud services (either capacity, software as a service, or platform as a service). So, I’ll take each of the three characteristics mentioned in Vogel’s blog and explain my view based on the fact that customers will make both economic and technical choices.
- eliminating costs – In reality there are data centers that work pretty well and are core to the business. The company has made an investment and therefore would not necessarily be able to lower costs. However, I expect that even if a company decided to go with a private cloud, there will be good reasons to use capacity on demand to fill gaps and expand for projects. In addition, a very large company will have the financial means to establish its own cloud that will be much more cost effective. A cost/benefit analysis of using a public cloud versus a private cloud is not straight forward. It requires a deep assessment of lots of different factors.
- elasticity – It is quite clear that many data centers do not have an efficient way to procure resources to users. However, if a data center is rearchitected to enable self-service provisioning, it can be transformed to better support users. Again, I expect that customers will take advantage of additional capacity or platform services even if they have private cloud services. This is especially true for companies where their computing infrastructure is the foundation of their business.
- removing undifferentiated services – This will really depend on whether the data center helps a company differentiate itself. There are definitely services that offer no value to the bottom line that should be placed in a public cloud (with a VPN for security, in some cases) such as electronic mail. However, where these services are at the core of the business and probably need to be in a private cloud. Many companies will select which services are not differentiated and which ones are and create a hybrid environment. Companies will have to do their homework both in terms of focus and costs. It might initially cost more to move a service such as email to a public cloud but will have huge resources in the long run. In other situations, paying per hour, etc. may be a lot more costly than you might imagine.
My bottom line is this. The cloud will continue to evolve over the coming decade and there is no one approach that will become the standard. The cloud is primarily an economic proposition that will require careful evaluation. Companies need to understand what their business is, what the value and role of the data center is and what is the best set of services available. The good news is that with the evolution of the cloud companies will have lots of good options.