Intel’s Otellini touts power of grids, green IT

Developing new products that balance computing performance and power consumption is one of Intel’s top priorities, according to CEO Paul Otellini. (Download this case study which details Intel’s current and future energy-saving initiatives to reduce costs and support business goals, and support the Climate Savers Computing Initiative, which aims to drive a 50 percent reduction in computer-related CO2 emissions worldwide. No registration required.)

The others are “socializing” enterprise networks, and supporting grid technology as a means of effectively channeling technology resources to distributed user communities.

All these themes featured prominently in Ottelini’s keynote at Oracle OpenWorld 2007 on Tuesday.

The Intel chief spoke of the sharp rise in the cost of power and cooling over the past few years.

“Data centre power consumption has doubled form 2000 to 2006,” he noted, adding that energy sucked up by data centres can electrify 5.8 million households for a year.

“This trend should not continue.”

Otellini said his company is keenly aware of the huge environmental and cost implications of excessive power consumption, which is why energy efficiency is a top priority with Intel.

“We take it very seriously.”

As part of this environmental commitment, he said, Intel is also implementing the industry’s first halogen-free products, and that’s a feature of the Intel Penryn processor launched Monday.

The new 45-nanometer (nm) Penryn for desktop PCs – the industry’s first lead free processor – is being promoted as a much faster, smaller and greener chip.

Otellini said to transition from 65nm to 45nm, Intel designed a new transistor and introduced technology that prevents electricity leaks, while improving energy efficiency.

This, some experts predict, is likely to result in a huge performance boost for desktops featuring the new chip.

For instance, Dean Freeman, an analyst at Gartner Inc. estimates Penryn will be 20 – 50 per cent faster than Intel’s previous chip releases in running general purpose applications (such as the Microsoft Office apps for instance) and 10 to 40 per cent faster in technical applications, multimedia and games.

Potential productivity benefits to business from the 45 nm Penryn were demoed by Cameron Purdy, vice-president of development, Oracle Coherence.

Purdy ran Oracle’s Coherence Data Grid application – that’s used on several Wall Street applications, reservation systems and e-commerce Web sites – on two sets of six servers, each set strung together in a data grid.

In the first set, each server was loaded with two Intel 65nm CPUs (Intel’s top performing chips before the launch of the 45 nm) with dual core processors.

Likewise, every machine in the second six-server set also had two CPUs, but these were the new quad core 45 nm chips. (Intel shrank the size of the transistors within these CPUs with a view to making them run faster, while consuming less power).

The standard number of financial transactions the 65 nm CPU grid was able to crunch was around 3 million per second.

By contrast the 45 nm system was able to crunch 30 per cent more transactions per second when just three of the six servers were provisioned for the grid.

“Essentially, we’re doing 30 per cent more work with half as much hardware,” Purdy noted. “What’s more we’re doing it with significantly less power and heat.”

He then demoed how – when the remaining 65 nm servers are thrown into the mix – transactions crunched per second go up from 3 to 8 million. And the icing on the cake, Purdy said, is that severs can be added to the grid without information loss or having to stop the application.

Another Oracle executive related how Intel servers are being used to run Oracle’s rapidly growing OnDemand business (Oracle apps delivered over the Web).

“The customer base for Oracle’s OnDemand has more than doubled since last year,” noted Juergen Rottler, executive vice-president, Oracle customer service. “This year we have 3.6 million end users compared with the 1.7 million we had last year.”

He said Intel processors are deployed as part of a virtualized environment on which the OnDemand business is run.

With more than 350 offerings, speed to market and the ability to grow quickly are both vital capabilities for Oracle OnDemand business, Rottler said.

“With [our] massive growth in this area there was the real risk we would run out of data centre space and power.”

He said “breakthroughs” generated from Oracle’s work with Intel at the chip level, as well on virtualization technologies have allowed Oracle OnDemand to double its business in the same data centre space – within a year.

Grid computing, Rottler said, is now integral to the way Oracle runs every aspect of its business.

For instance, he said this technology is used at Oracle University – Oracle’s online training program that instructs 350,000 students a year, in 56 countries and 24 languages.

“It’s a good-sized operation representative of the businesses that many of you are in today,” he told the audience.

Oracle University, he said, was faced with many of the same challenges as the OnDemand business.

“So last year we re-architected the grid [structure] that business runs on, utilizing several hundred Intel Xeon servers.”

In doing so, he said, Oracle University business has been able to drive down its total server count by more than 70 per cent, decrease space utilization by 52 per cent and power consumption by 47 per cent.

“Most importantly, we can re-provision our global ‘classrooms’ on a daily basis. We can re-provision servers overnight on our grid. We’ve been able to add a new class to our curriculum every day, and do that with the same [data centre] capacity and these dramatically reduced numbers.”

Intel’s Otellini said such amazing results are possible largely because of the tremendous improvements in server processing power over the past three years.

For instance, he said, in 2004 it would take six racks of 126 servers occupying 240 sq ft and consuming 48 kw of power to generate 5.1M bops (billion ops per second).

“Today you can replace all those machines with a single rack [containing] 17 blades, readily deployed with air cooling. And it can be done with an 83 per cent reduction in floor space, an 87 per cent energy reduction on the same workload, and annual savings of $53,000 a year.”

Even more dramatic improvements are in the pipeline, he said as Intel intends to bring more processing power to bear on the issue of performance vs. power.

Related Download
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center Sponsor: Lenovo
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center
Find out how Hyperconverged systems can help you meet the challenges of the modern IT department. Click here to find out more.
Register Now