GRIDtoday Logo AMD

DAILY NEWS AND INFORMATION FOR THE GLOBAL GRID COMMUNITY /

   ( Table of Contents )   

Special Features:

NPF LAUNCHES NETWORK COMPUTING INITIATIVE

At the Communications Design Conference (CDC) in San Francisco, The Network Processing Forum (NPF) announced the launch of a new study group -- the Network Computing Initiative (NCI). This new group will work to evaluate how network processing technologies can be used to further increase the performance and efficiency of networking components in the data center and throughout the network. Noting recent industry trends such as the offloading of advanced networking functions in multi-tier or scaled-out data center frameworks and the growing need for multi-location deployment of advanced security services throughout the data center and the network, the group will examine what interfaces and architectures are needed to address these areas with multi-vendor interoperable network processing solutions. Details of the initiative will be presented at 5:30 p.m at the CDC NPF panel session on Wednesday, March 31.

"This new effort will expand the scope of the NPF and attract data center players to participate in the Forum," said Lars Petersen, board member of the NPF. "Our goal is to facilitate the development and adoption of solutions that utilize network processing technologies. Establishing industry accepted interface specifications will lower development costs and speed these new products to market."


Web services, utility computing, .NET, CPU harvesting and distributed computing are just a few of the technologies that fall under the Grid computing umbrella. Gt04 -- a premiere enterprise Grid computing conference targeting industrial and commercial users -- will gather experts, and outline strategies and road maps for Grid deployment. For more information, visit www.gt04.com.

Grid computing is here!


Scalable performance and dependability are vital to the success of today's data centers. A common technique for achieving these two objectives is to use a multi-tier framework or scale-out model. This model groups server resources such as individual servers or blades into a cluster configuration and uses a layer 4 and/or layer 7 switch to create a single Virtual server. This approach increases scalability and dependability by enabling the adding or removing of physical servers/blades without changing the network topology. It also enables the ability to automatically switch to, or provision another server/blade for immediate capacity increases and provides the capability to mask planned and unplanned outages of an individual server by shifting its workload to another server. One of the things that the study group will look at is what kind of network processing interfaces must be developed to enable the use of network processing technologies to maintain this highly scalable and dependable model while reducing components costs, increasing efficiencies and enhancing overall performance.

Another key area of application to be investigated by the study group is the development of interfaces and protocols that enable multiple advanced network services to be easily deployed throughout the network on multi-vendor network processing-based platforms. For example, many of today's security threats are more sophisticated and require solutions that not only monitor header and flow classification, but are also aware of a growing number of control protocols and perform content processing for intrusion detection or virus detection. To be effective, these capabilities must be deployed at multiple locations throughout the network and have the performance capabilities needed to analyze, detect and prevent these threats at wire-speed.

The study group will initially create a functional architecture and model several network computing use cases to understand the various interactions between components. This examination will include a study of the problem from multiple perspectives including chip to chip, module to module, plane to plane and function to function. The group will take steps to avoid re-inventing existing interfaces as defined by relevant groups such as the IETF and others. The result of this exercise will be a proposal for a new set of interfaces. If approved, the NPF will form working groups to create the new interface specifications.

"The lack of standard API's that enable multiple applications to execute on a single hardware platform has been a key factor in holding back the proliferation of network processing technologies into these types of applications," said Steve Hunter, chair of the NCI study group. "This effort will attempt to evaluate how the industry can better enable the application of network processing in server architectures and data center environments. The group will also investigate how best to enable the portability and integration of today's and tomorrow's advanced network applications."

( Top of Page )

   ( Table of Contents )