Cloud Computing in the Real World: Choosing the Right Cloud Model
Our pros and cons list will help you sort out the issues relevant to your situation and select the right cloud model -- public, private, or hybrid -- for your environment.
By David S. Linthicum, Founder and CTO, Blue Mountain Labs
Private, hybrid, and public clouds are the models of cloud computing put forth by NIST a few years ago. Today, these are the largely accepted models. However, many sub-categories are beginning to emerge that blur the lines between private, hybrid, and public models. How do you choose? It’s best to start with a good understanding of each model.
Public clouds can be leveraged by anyone who would like to become a public cloud customer. Amazon Web Services (AWS) and Salesforce.com customers use shared infrastructure and applications delivered over the open Internet. You can provision or de-provision the resources on-demand, only paying for the resources you use.
Private clouds mean that you run infrastructure that mimics a public cloud, but does so on hardware and software that you own and maintain in your data center. Many cite security and privacy issues with public cloud computing providers when they opt for private clouds. In certain instances this is a valid excuse, but many select private clouds because they can check “cloud computing” off the technology to-do list and thus appear to be innovative while they maintain control.
Hybrid clouds are a mix of both public and private clouds. They allow processing loads to be shared between systems within your data center and public cloud providers, perhaps even moving processing and storage between public and private cloud instances dynamically. There are many companies emerging that plan to provide operating and management systems for hybrid cloud computing environments as enterprises and government seek to create systems that leverage the best of both private and public clouds.
Let’s put the hype aside for now and look at these approaches in the real world.
- Typically the less expensive option considering that you only pay for the resources you use. There are no hardware and software maintenance costs.
- Provides huge scalability; you can allocate as many server and storage instances as you need.
- Outages can occur.
- Network latency issues can arise because you typically send huge amounts of data back and forth over the open Internet.
- Security can be a concern, and there is a risk of data compromise.
- You own the hardware and software, so security is in your own hands.
- Performance is typically better because the systems are communicating with the end users and clients over private networks.
- You lose the cost and value benefits of cloud computing because you’re really just implementing another internal system.
- Private cloud computing software is considered at the early stages now and is still evolving.
- Provides the best of both worlds, allowing you to balance system loading between private and public cloud instances.
- Technology is moving quickly in this direction means better tools are emerging.
- Moving images, processing, and storage between private and public cloud instances is not as easy as you may think. Technology has some growing to do.
- Could cost more than either the private or public cloud computing options considering the technology and skills required.
A Final Word
There are no simple choices here. You need to size up the requirements, consider the technology available and the business case, and then map the right solution to those requirements. If that sounds like what you’ve done for years to make platform and infrastructure selections, you’re exactly right. However, you also need to learn a whole new set of technologies, and that’s the true challenge.
David S. Linthicum is the founder and the CTO of Blue Mountain Labs and an internationally recognized industry expert and thought leader. He is the author or co-author of 13 books on computing, including Enterprise Application Integration (Addison Wesley). David is the keynoter at the 37th International Computer Measurement Group's annual conference (CMG'11), December 5-9 at the Gaylord National Hotel in the Washington, DC area. You can contact the author at firstname.lastname@example.org.