From a commercial point of view grid is moving in two directions: SOA and virtualisation. The marketing messages that used to deal with grid or even web services, have been rewritten to speak about SOA and/or virtualisation. Even traditional grid companies (such as Data Synapse ) now define themselves as providers of application virtualisation software.
Service-oriented architecture (SOA) is a software architecture based on loosely coupled components each a separate software service - in order to deliver a solution to the user based on the requirements of the business process. Each of the network resources is made available in such a way that they can be used without needing knowledge of the underlying platform.
Virtualisation is essentially the hiding of technical detail, through a process known as encapsulation. The external interface virtualisation provides can be used to simplify control systems, for example, or to combine resources at different physical locations.
Virtualisation is any abstraction layer that permits aggregation, emulation or partitioning. Clustering and grids, for example, are abstraction layers that permit aggregation.[1]
Effectively the application doesnt see the networked machines it is run on as separate servers, rather as a single continuous machine. The emulation layer lets software designed for one system run on another and the partitioning layer divides the single virtual resource so that each running programme acts as though it were running on a single server.
For more information see:
www.the451group.com/special_reports/special_report_detail.php [1] and chucksblog.typepad.com/chucks_blog/2006/12/grid_and_virtua.htm
Grid technology is moving into the arena of transactional applications. Because transactional applications such as e-commerce applications, trading, and real-time analytics generally demand 100 percent availability and are seeing exponential growth in the transaction workload, there are several initiatives to study ways to integrate Grid technology and transactional applications support.
Platform Computing and GigaSpaces announced (Jan 2007) a partnership to make transactional applications work over a compute grid. Other vendors trying to exploit this different approach are Appistry, Paremus, Aumega Networks and Majitek.
The idea of bringing together grid resource virtualisation and management solutions and the end-to-end linear scalability of transactional and data-intensive applications will create a single platform capable of addressing the data, messaging and processing bottlenecks present in high-throughput, low-latency transactional and data-intensive applications.
From discount travel sites to airline company sites or shopping sites like Amazon, businesses need to reduce the cost of processing each transaction. Grid platforms with their ability to utilize commodity hardware can bring a solution to these businesses.
This material is taken from the document "Market Watch" produced by the BEinGRID project.
Print this page