The liquid-cooled solution promises lower operating costs, better efficiency, silent operation, increased reliability, and a smaller economic footprint.
During the Gartner IT Infrastructure, Operations, and Cloud Strategies Conference, Schneider Electric announced a partnership with Avnet and Iceotope that it says will result in the industry's first commercially available integrated server rack with chassis-based immersive liquid cooling.
The new system will be able to operate with lower costs, increased efficiency, better reliability, and be more economically friendly, said Kevin Brown, senior vice president and CTO of the secure power division of Schneider Electric.
Using Iceotope's liquid cooling technology and a high-powered GPU, Avnet helps integrate the liquid-cooled server with Schneider Electric's NetShelter liquid-cooled enclosure system to deploy into data centers or edge computing environments, Brown said.
SEE: Special report: A guide to data center automation (free PDF) (TechRepublic)
"We're trying to bring a new way of doing liquid cooling into the market, and we're driving out a longer-term opportunity of bringing energy consumption down in data centers and building a much more resilient edge," Brown said.
Why liquid cooling?
"Liquid cooling has been around for a very long time; it's not a new technology," Brown said. "Our interest on it started increasing within the last 18 months to two years because there's a trend occurring with GPUs and CPUs, where the amount of power that they're consuming is going up."
"For years, Intel CPUs only consumed about 150 watts," he said. "If you look at the latest chips now, whether it's the GPUs or latest CPUs from Intel, they're consuming more like 300 to 400 watts. That presents a challenge for air cooling."
Because air cooling cannot keep up with high-compute processing that companies are doing, liquid cooling is the most-effective option, Brown noted.
There are two primary forms of liquid cooling: Direct-to-chip and fully-immersed.
With direct-to-chip, "you'd take a pipe with water and you run it to the chip, then you run it out, and it just runs through a little heat exchanger. With fully immersed, you take the servers and put them in a big tank of oil, like mineral oil," Brown said.
However, both tactics present their own risks. With the direct-to-chip cooling, organizations run the risk of having water leak inside the IT. To prevent this, many companies integrate a leak prevention system, but those are costly and use a lot of energy.
This strategy also doesn't always capture all of the heat from the servers, requiring additional fans and airflow, according to Brown.
The main disadvantages of the oil-based immersive cooling involve the mess. An oil bath is extremely messy and introduces concerns over flammability, he added.
How is the chassis-based immersive cooling rack different?
With the chassis-based immersive cooling system, Schneider Electric hopes to merge the positive aspects of both direct-to-chip and tub immersive strategies, while eliminating the disadvantages.
By bringing the standard rack form factor of direct to chip, but comprehensive heat capture of tub immersive, organizations get a cost-effective, high efficiency, reliable, fan-free cooling system, according to Brown.
"What we're doing is trying to take some of the convenience that direct-to-chip brings with some of the performance that you can get out of immerse solutions," Brown said.
"That's why we like to call it rack-based immersive. It's a little bit different than just immersive, but it's all the benefits of immersive with convenience and the footprint of a direct-to-chip. And, nobody's done a rack-based solution in the past."
The benefit of this system is that the company isn't forced to have a room of big tubs filled with oil. Instead, companies receive a standard-looking IT rack that they can drop in their data centers.
The immersive cooling technology performs neatly right inside the rack, said Steven Carlini, vice president of innovation and data center of the secure power division of Schneider Electric.
The system is useful for any company in any industry that is running high-compute intensive applications, Brown said, such as artificial intelligence (AI), machine learning, and big data analytics.
"The main one you're seeing is an AI application, specifically AI learning applications. Those draw the most power, and the more power the servers draw, the more heat they give off," Carlini said.
"You're also seeing things like blockchain, where those are very, very compute-intensive. Any compute-intensive applications that's requiring these graphic processor units, these Nvidia chips and these next-generation Intel chips are really what's driving it," Carlini said
"So you're seeing a lot more data being created because of IOT and certain applications and you see the need to run all of these, process these with GPUs. So any of those applications that are very processor-intensive are really driving the need for liquid cooling.
Schneider is shooting for availability in Q1 of 2020, with availability for select customers in select regions. However, wide availability should be available in the next 12 months, Brown said.
For more, check out Schneider Electric: First industrial smart factory in the US opens in Lexington, KY on TechRepublic.
- How to become a database administrator: A cheat sheet (TechRepublic)
- 10 things companies are keeping in their own data centers (TechRepublic download)
- Hiring kit: Database administrator (TechRepublic Premium)
- How hyperscale data centers are reshaping all of IT (ZDNet)
- Best cloud services for small businesses (CNET)
- DevOps: More must-read coverage (TechRepublic on Flipboard)