EDCi - Cloud Computing

Cloud computing is a means of providing services or applications to many users via a remote connection. Depending on the application and the expected number of users, cloud computing infrastructure can mean anything from a single Microsoft Sharepoint server to a cluster containing thousands of servers for application providers like Google, Salesforce.com, Amazon or Yahoo.

The concept of cloud computing is fairly simple: build and maintain one pool of resources and make it available to users everywhere via a standard Web browser. There could be one centralized resource pool to manage and maintain, or, to guarantee high QoS during peak demands, there could be multiple resources in different locations connected together via high bandwidth connections.

Cloud computing is emerging as a strategic model for large IT organizations to outsource some of their IT needs and also to gain access to needed resources in peak demand, but the data center that supports the cloud can deliver poor results unless it is optimized for performance, efficiency, scalability and robust QoS.

The Efficient Data Center Initiative provides best practices for an optimized cloud computing data center:

Performance - Create a balanced data center with appropriate processing power, connectivity bandwidth, and storage performance to eliminate bottlenecks. Elements include:

  • Multi-core, multi-processor servers in a clustered architecture that allows multiple servers to share the workload of providing an application or service.
  • Virtualized storage pools that simplify storage management
  • Data center virtualization to simplify server provisioning
  • High-bandwidth, low-latency interconnect among servers and storage to eliminate the network as a processing bottleneck.

Efficiency - Minimize CapEx and OpEx by using less equipment, management costs, and lower power consumption

  • Deploy virtualization technology to make the most of server performance while reducing power consumption
  • Consolidate I/O adapters to reduce management costs, capital costs, and power consumption
  • Use high performance computing solutions to ensure a good user experience, reduce response time and be able to host maximum number of users in the cloud

Security - Protection of the cloud resources, the transmission media connecting them to end users and the between the users application in the cloud.

Quality of Service and Service Level Agreements - Cloud computing will fail as a model unless the cloud can deliver acceptable quality of service and adhere strictly to service level agreements. To maintain high quality of service, data centers should adopt the following elements:

  • High compute, storage, and I/O performance that is balanced across the data center and is free of bottlenecks that cause servers to delay processing.
  • The lowest possible latency in connectivity and I/O systems