The Haziness in Cloud Computing
Cloud Computing represents both a fundamental and universal change from a desktop- to network-centric software model. The key to understanding cloud computing is that its infrastructure — network, storage and other services – is abstracted away from users to point that it “just doesn’t matter.” You merely use a browser to access these services and data.
To put it simply: the Cloud is the computer.
The iPhone’s application delivery model has ushered in a new example and heralds the Cloud Computing era to come. A few key applications are pre-installed. Instead, a wide variety of applications are easily downloaded on the phone, data resides in the Cloud, and data synchronization transforms conventional desktop applications (like Exchange) into a temporary utility server until everything is up in the Cloud.
The IT companies that will succeed will embrace this new, Internet-centric model, but not merely by sprinkling “Cloud Dust” — like fairy dust — on their offerings in order to benefit from this significant wave of interest by corporations, VC, and the general market.
“Cloud Dust” is contributing to the haziness handing over Cloud Computing, in the same way that free and open source software experienced haziness in 2004 and 2005 before the market arrived at its tipping point.
Hosting companies, Software as a Service (SaaS), entrepreneurs, and others are actively repositioning, expanding definitions, developing new features, solutions and support, and doing all manner of gymnastics to catch and ride the Cloud wave.
Virtualization technologies and hypervisors have enabled data centers for Cloud Computing services. Open source software has also played a pivotal role by being the foundation for today’s Cloud Computing infrastructure. The reliability and fast deployability of Linux, Apache, etc. has empowered this shift, along with the change of computing habits by the Joe and Jessica End User. With 75 million Facebook users and 40 million Twitter users, Cloud Computing has a basis for success in end-userland.
The basic taxonomy of Cloud Computing can be summarized in three services:
- Infrastructure as a service (IaaS), which includes large vendors — especially Amazon’s EC2, RackSpace, and soon others;
- Platform as a service (PaaS) like Google, Microsoft, Bungee Connect, and others (“If you can program it, then it’s a platform. If you can’t, then it’s not”); and
- SaaS vendors – especially Google Apps, but also including SalesForce.com, Skype, WebEx, and others.
Still, questions abound: Are private Clouds – mainly from corporate citizens – included in IaaS? What is the difference between EC2 and a bunch of HP or Sun boxes with a hypervisor?
Cloud Computing is going through a period of rapid experimentation, adoption, and repositioning. New entrants into the market – like the California Gold Rush of the 1840’s, or the Internet Bubble of the 1990s — add more intensity and noise to an already chaotic field. This haziness, or chaos, of Cloud Computing will persist for the foreseeable future, until the tipping point is reached.
Tags: Cloud Computing
September 9th, 2008 at 7:46 pm
[...] The Haziness in Cloud Computing [...]
October 7th, 2008 at 1:33 pm
Are private Clouds – mainly from corporate citizens – included in IaaS?
Absolutely. That’s the whole point! Corporate citizens buying infrastructure as a service. As an IaaS provider, I’m seeing this trend progress significantly.
What is the difference between EC2 and a bunch of HP or Sun boxes with a hypervisor?
The API mostly I suppose. EC2 got the API right first. Now, others are following. Xen and VMWare have had some semblance of an API for ages. But, it took Amazon and their published EC2 API to really make it fly.
A agree that the cloud is the computer. That’s more or less what I was saying when I wrote http://www.productionscale.com/home/2008/8/21/cloud-computing-back-down-to-earth.html
What I think is most interesting is that it’s Software and the developers that write it that are having a difficult transition to the cloud. Most of the applications that people use and are writing today do not run well in a cloud or the just don’t really run all that much better. You still get some benefits to be sure, but not the TRUE benefits. These true benefits require people to essentially adopt many of the practices of the HPC world along the lines of distributed parallel programming methodologies.
I wrote several things a few months ago as well that have much in common with your thoughts here. I would be interesting to compare notes.
http://www.productionscale.com/home/category/cloud-computing
Thanks for an interesting article.
-Kent