Networking vendor D-Link has published tests that appear to back up the startling power-saving claims the company has been making for its range of 'green' switches.

Commissioned from independent testing outfit, Broadband-Testing Labs, the test report pitted two Gigabit Ethernet Layer 3 switches from the company's standard xStack range against the company's more recent Green Ethernet DGS-1224T, to see which was the most power hungry.

The tests found that the 'green' box used considerably less electricity than the non-green products in a variety of scenarios, including with ports active but not passing data, at 50 percent load, and at maximum load.

In an idle state, the 1224T used 20.7 watts compared to 32.2 watts for the non-green xStack DGS-3627, and 33.7 watts for the DGS-3426P. As the switches were loaded at 50 and 100 percent, the 1224T increased its power consumption by only a small amount to 23.6 watts at full load, well below the 3426's 46 watts and 3627's 50.3 watts.

Importantly, the consumption tests were conducted under what Broadband Testing describes as real-world traffic conditions, necessary because it allowed protocol loads for http, FTP, and video to be simulated using a specialised testbed system. This gets round the criticism often levelled at vendor power claims that tests are not conducted under conditions that model how the products will actually be used by customers.

"Not only are the consumption figures well below any published figures currently available, but are below those D-Link itself made available from internal testing," said Broadband-Testing Lab head, Steve Broadhead. "D-Link was first to market with Green Ethernet products and made claims about energy-efficiency that have been validated by these tests," he said.

D-Link puts the superior performance of its green switch down to a number of design features, such as lower chipset power drain, the ability of the switch to turn off its cooling fan, and being able to detect when a port is not connected to an active device such as a PC. If the latter applies, the port can be turned off.

Power consumption is said by D-Link to be so important to its customers that plans to roll out the power-saving feature to all products, business and consumer, in the next "six to nine months."

Interesting though these figures are, no vendor has yet dared to compare its 'green' products with those from a rival vendor, and for a simple reason - there are no universally-agreed tests which can be used to carry out comparative testing in a way that would be fair.

"Vendors own published power consumption figures are meaningless - we have found them to be very inaccurate during internal tests in our labs," the official Broadband-Testing report notes.

It is believed that Broadband-Testing Labs has been asked by D-Link to look into the possibility of developing such tests, but it is hard to see why vendors would submit products if they suspected they might not score highly on them. Until such tests come to pass, companies will have to score power consumption using their own judgement of vendor claims, or the odd test graphs that do emerge from time to time.

However, what the latest tests do suggest is that power consumption for networking kit can be lowered by a considerable margin, compared to what passes as normal power drain for most of today's equipment.