Characteristics of Cloud Computing

  • Scale: A cloud compute platform should scale both vertically and horizontally. Vertical scale implies an on demand reconfiguration of the underlying machine (memory, cpu, network, storage). Horizontal scale means the number of such machines or instances which are dedicated to running your software.
  • Utility Based Flexible Pricing: Your vendor should charge only for the resources consumed (cpu, memory, network, storage, others).

 

 

 

Did you know?

Migrating your app to the cloud does not guarantee that your TCO is lowered and ROI improved. Various factors like the architecture of software and its implementation still get in the way of performance and scale.

A poorly architected app will only increase your costs. Contact us to alleviate your app performance problems.

The 3 Strategies

Review these three simple suggestions to get an idea of what your strategy should be.

#1  As Is Deployment

Move your application from your current datacenter to a cloud datacenter like Amazon. Rent Amazon EC2 + EBS instances to run app and store data respectively. This is a very simple, yet significant, step towards cloud adoption. While the management of the instance is still your responsibility, you can enjoy the benefits of vertical scale to lower TCO.

Amazon EC2 pricing starts at $0.085/CPU hour for Linux. Windows and full pricing at http://aws.amazon.com/ec2/pricing/.

This option is recommended for most existing applications as it offers the most flexibility and run time compatibility for your applications.

#2  Managed RDBMS Deployment

To solve a major IT headache, i.e. data protection and availability, many cloud vendors offer a managed RDBMS (Database) solution. They take up the ownership of keeping your database instance up and running with specific SLAs.

Amazon RDSAzure Tables, Google Hosted SQL all fall in this category.

#3 Application Re-Write

Individual instances have limits on how much processing they can do in a second, and stop scaling after that limit has been reached. Distributed computing techniques go beyond these limits and can scale to millions of requests.

Scaling CPU has been affordable to most use cases for a number of years now. However scaling the storage has not been that easy. Distributed datastores like Google's Datastore and Amazon's SimpleDB offer great programming flexibility and performance/scale.