Grid computing on .Net is now possible due to the efforts of Oakland, Calif.-based Digipede Technologies. The company offers the only parallel processing solution that runs .Net code. Digipede is built entirely on .Net and unlike its Linux counterparts, the solution is extremely easy to set up and manage.
By using agents that can be installed on any commodity hardware such as PCs and servers, the Digipede software can distribute code objects within programs across a network of machines and execute them remotely. This mechanism is analogous to how Windows executes COM objects. DOS programs and COM objects can be distributed on Digipede's network as well.
By taking advantage of grid computing programs, .Net code can achieve the most efficient execution of encapsulated, nonsequential tasks. Nothing in the market really competes against grid computingnot even enterprise technologies such as Java application servers, which cannot easily match its scaling capabilities.
While parallel processing has been around for years in the scientific community, primarily the financial sector has taken advantage of it to run complex business analysis. However, analytical applications such as portfolio optimizations and Monte Carlo simulations are not the only applications that can benefit from grid computing. Transactional applications also are ideal for grid computing and can be divided up across processors. For instance, billing applications can execute on a grid quite easily. By separating data into chunks and independently managing these resources remotely through Digipede agents, encapsulated objects from a billing system can manage processing and even simultaneously print reports on multiple printers without communicating with other code instances or the master instance.
In addition to providing higher throughput, Digipede has removed the intimidating factor associated with buying grid computing software, installing it and inserting code into existing .Net applications. Digipede software provides multiprocessing execution, coordination of resources, application load balancing and data management services that controls data access and movement around a network.
Developers only need to extend and reference Digipede objects and insert events in the right places to make the application parallel. Instead of throwing everything into a grid, developers also can create subclasses that include relevant information for calculations that can execute in parallel. In a matter of 20 minutes, .Net developers can add Digipede code into their programs. All program dependencies, including data access, is managed automatically by the Digipede server.
The company's three-tier channel program is broken down by yearly sales goals (50,000 of Digipede software per year) and prospective partners must commit to having at least one representative attend a Startup Digipede Training course, with a fee of $500/year (waived the first year). Digipede offers partners a 15 percent discount on the first $50,000 of Digipede software, maintenance and premium support sales (per year). Those discounts increase with sales volume.