In my earlier post, I gave a short review on the Animoto iPhone app. Animoto is a classic example of why cloud computing works best for them. Animoto was first made available on Facebook in April and within a week, the company had to increase the number of servers with Amazon's Elastic Computer Cloud (EC2) from 50 to 5,000. To a startup company, it is going to be overly capital intensive to invest in so many servers. Besides, It will also be a hasty decision for the company to buy the servers and is also tough for the company to get the servers running up quickly due to the set up and installation required. An opex model where company only pays for what it uses will be a better business model for a start up like Animoto until it establishes its baseline of usage.
There is a recent good article by Computerworld that talks about cloud computing. I will highlight the key points, coupled with my own views.
1. As illustrated in the Aminoto's case, cloud computing can bring savings to a start up who has yet to establish a usage baseline. Cloud computing also offers scalability and redundancy. One point to note is that it's easier to scale up but the company using cloud services need to be more careful when scaling back as there can be some residual processing resident in the servers that are being switched off and this may cause disruption to the service. The company needs to do the proper clean up first before scaling back.
2. Cloud computing providers need to create different types of pricing models so that it can cater to different needs. For example, the article suggests that for data that has not been accessed in more than 6 months, the provider can move to a slower form of media and charge a lower price for it. The view is that it's not economical to store data in the cloud over a long period of time. Shop around and choose one that best meets the needs but not sacrificing future scalability (we all think big right?!) Check if the company provides detailed billing or audit report for verification, if the need arises.
3. Cloud computing is becoming popular but the performance may not be uniform across various geographies. Getting SLGs from cloud computing provider is important. In my view, it is also equally important to understand the experience of the cloud computing providers. What is their expertise and their main clientele base? What is the customer base's geographical composition? What is the scalability and redundancy plans and where are their data centres. Another important note is to understand their bandwidth upgrade policies.
4. Have back up servers for the mission-critical applications in the event of a service disruption. The idea is not to have your business totally disrupted if there is a complete outage of the cloud. If you can't afford the servers, perhaps you should diversify your requirements across multiple vendors.
5. If you are subscribing to an application provider that is running the application on a cloud, find out whether it is a native 'cloud' application or it is a thick client app that is made to fit into the cloud. If it is the latter, understand what are the restrictions between a 'cloud' app and a thick client version.
Courtesy of Computerworld, the chart below also shows you the potential of IT cloud services are likely to be in the area of collaborative apps, IT Management apps and Personal / Biz Apps. THe survey participants characterised current an future usage - on a scale from 1 (none) to 5 (widespread) - of a variety of IT cloud services within their organization. The chart shows the percentage of respondents toward the higher end (4 or 5) of the usage scale now and / or who will be there in 3 years.
Just in case you are wondering what is cloud computing or what is cloud computing isn't. Computerworld article lists down 4 different styles of cloud computing offerings:
- Software as a Service (SaaS): special-purpose software made available by a third party over the internet with usage-based pricing model e.g. Salesforce.com, Zoho (my own personal view is that SaaS is not cloud computing but rather a form of cloud service).
- Platform as a Service (PaaS): An integrated software environment for whcih systems administrators and developers can build, test and deploy custom applications. E.g. Google Apps Engine
- Infrastructure as a Service (IAAS): A service that providers the core computing resources and network fabric for the cloud deployment e.g. Amazon's Elastic Compute Cloud
- Core Cloud Services - standalone components built on cloud platforms that can be woven into cloud applications, such as billing systems management and storage e.g. Microsoft BizTalk, Right Scale.
Here's another article that tells you what cloud computing isn't.
Posts
No comments:
Post a Comment