Asby CNET News.com, the N1 project is an attempt to "virtualize" computing equipment, making servers and storage look like giant pools of resources with computing processes swimming within.
Sun Chief Technology Officer Greg Papadopoulos unveiled the concept Thursday at the company'shere.
Essentially, N1 describes what happens when a network of computers and storage systems is assembled into a much larger whole. "This is just like what an operating system did for a computer over the last 20 years. It's all the stuff we did in Unix and redid in Linux," he said.
The concept resembles efforts under way at some of Sun's biggest competitors. IBM is working on "" computing that unites servers and storage into a large pool. Grid computing grew out of academia, though, and focuses more on mathematical calculations than the business processes at the heart of N1.
planetary computing initiative similar to the grid efforts.also is a fan of grid computing, and Sun itself has released its own as an open-source project. Meanwhile, Hewlett-Packard has its own
Research group IDC terms the concept behind N1 "service-centric infrastructure," said analyst Vernon Turner. "You won't organize boxes anymore, you'll organize resources."
Papadopoulos said the trend is starting in storage. Some computing jobs require lots of storage, but others require little. When each server comes with its own directly attached storage system, some space goes unused, depending on the job. Through virtualization, storage space is pooled so its storage can be used more efficiently.
The next stage after storage will be to create a virtual world for computing resources, Papadopoulos said. In vogue today is a combination of large and small servers for data centers. The small, inexpensive "edge" servers handle chores such as sending Web pages or audio streams. And in the heart of the data center are large, expensive symmetric multiprocessing servers with many CPUs (central processing units) and high-end features.
N1 will understand all the components of this data center, Papadopoulos said. With it, administrators will be able to allocate more computing power to important tasks and better manage less important ones, a process similar in concept to the way high-end Unix servers and mainframes can be partitioned so they can run several jobs with shifting priorities.
But to fulfill the promise of N1, Sun will have to deal with computing hardware from other companies, Turner said. It's likely computing processes themselves will run using Sun's Java software and the industry-favored XML (Extensible Markup Language) data description language, because those technologies work well across different companies' products. But a company's data center is likely to be populated with many different companies' hardware.
Sun is well aware of this reality. Java runs on any major server, while Sun is working to spread its Sun One software plan to. Additionally, the company sells software that enables sophisticated data-protection features that work with any storage system.
N1 has the potential to work on mammoth computing systems, Papadopoulos said. "This doesn't stop at hundreds of processors," he said, but rather spans all the way up to tens of thousands of CPUs as well as petabytes and exabytes of storage.
These mammoth systems will be needed to match the computing demand imposed by all the network-enabled devices Sun expects.
There are about 100 million desktop computers and servers in existence, but Papadopoulus expects about 100 billion Internet-enabled cell phones, cars, appliances and other devices soon, with about 100 trillion Internet-enabled thermostats, mail packages, articles of clothing and other devices in the more remote future.