The what, where, when and how of the cloud are persistent questions that must be answered correctly if a cloud deployment is to be successful. But, misconceptions can be handicaps and organisations often labour under a number of misconceptions. A little understanding of cloud architectures, management and chargeback can be useful in selecting the best fit solution to their needs.
Managing and using different cloud architectures
The cloud has evolved from the convergence of a number of technologies and approaches to computing. The underlying architecture is similar to and different from existing computing models and impacts on the operational and technological approaches to network configurations and security practises. Like all computing systems operating over a network, the cloud consists of a back end [the remote server(s)] and a front end (the client computers). The connecting network is the Internet. The servers, the applications and the storage devices at the backend provide a cloud of services to the customers. Cloud computing systems that cater to multiple clients are known as “public” clouds. When an entire cloud service system is dedicated to a single client, it is known as a “private” cloud. Hybrid clouds combine features of the public and private clouds.
The client machines connect to the remote server(s) and the applications using software called an “agent”. The agent is a special kind of software, known as middleware. It enables IT Administrators monitor traffic, administer the system and set rules and regulations for access and use of the information stores available in the remote server.
“Utility computing” is the unique selling point (USP) of the cloud. Organisations signing up for cloud services agree that the cloud makes it easier for the organisation to track and measure IT expenses per business unit. Chargeback becomes simpler as it is metered like electricity on a “pay per use” basis.
Chargeback mechanisms in the cloud take into consideration two factors:
What are the resources and metrics for chargeback?
How to account for excess capacity that is supplied on the fly?
The chargeback system is built on the assumption that customers tend to use average capacity rather than large capacity and hence offering scalable services does not automatically result in extensive usage of resources. Further, cloud vendors understand that successful chargeback systems separate infrastructure costs from service costs and that shared infrastructure is a combination of fixed and variable costs in which the percentage of fixed costs will decrease as number of users increase. Pricing will consequently, be, unit tiered; bundled or pay per use.
