Forget about smart use of air (hot-cold isles) or water (airco) for cooling down your intensively used servers. Just dunk them in a bath of mineral oil and you’re good to go! At first glance, that seems to be a possible conclusion after a thorough experiment by Intel. It’s not as strange an experiment as it may initially seem. The idea stems from a practice already used for industrial transformers, circuit breakers, capacitors and the like. So, it’s not a new thought at all.
The experiment took place for a period of a year and they compared the need of additional energy use needed to cool servers in a more ‘traditional’ way with similar servers being cooled by having them immerged in mineral oil. For it, Intel worked closely together with Austin, Texas based company Green Revolution Cooling. The result of this particular experiment?
- 60 % less energy used for cooling
- No need for cooling fans
- No intensive modifications to servers needed
- 10 – 20% cheaper compared to a high-end containment system
I wondered what the maintenance of such servers would look like. Coming home from work covered in mineral oil every day? Although this could mean the average Data Center expert might look like they have spectacular looking skin, it would seem quite a nuisance to have oil residue sticking to everything you touch. Although the article doesn’t really go in-depth to this aspect, it looks like a smart way to prevent such nuisances was taken into account. However, it does explain that one of the main conclusions taken from the experiment is that the unusualness of the idea is thought to be the biggest hurdle in accepting and developing this as a real and ‘best’ practice. But is that the only thing to consider?
Greener, but is it sustainable?
Besides the basics I learned about chemistry in secondary school and things I picked up here and there throughout the years I am not an expert at it. Though the results seems to make it a possible attractive alternative to the conventional cooling methods, there’s one thing that may eventually put a spoke in the proverbial wheels. Intel’s partner states on its website that it’s eco-friendly and non-toxic, however I could not find information about the argumentation behind that statement. Just a few clicks and wiki was reminding me that mineral oil is made out of petroleum. This tempered my initial enthusiasm. Not diverging into other possible objections, this alone would mean it will not be sustainable in the long run. If it were possible to do the same using ‘essential oils’ you might think this would be better. There’s one big but here too though. I believe the amount of this oil needed to be able to practice this on a large scale could potentially be a competitor to food related agriculture or a driver in deforestation which of course is not sustainable either. Perhaps if it was an oil derived from a plant that grows on places not suitable for food or forests, but the question would still be whether that would be able to sufficiently meet demand supplies.
Nonetheless, this experiment may just prove to be the start of a process towards a sustainable variation in the future. So, as long as we see it’s current version as an intermediate step – a less energyintensive alternative for now, it still made it an interesting read in my books.
See link below to read the article that triggered this blog. Click on the image to go to the GRC gallery page.