back to article Don't waste your energy on Docker, it says here – wait, that can't be right...

Docker may be the darling of DevOps, but it's something of a minor extravagance when it comes to energy consumption. Computer science boffins affiliated with the University of Alberta in Canada have spent some time comparing the energy used running software applications in Docker containers with power consumption on bare metal …

Silver badge

Business Evolution

Everyone's bit barn costs is people, to begin with, when the business is small. When the business scales up, it starts being energy, big time...

So I'd say that eventual business size has to be considered too. Plan that scale up in the beginning.

Facebook and Twitter didn't plan a scale up, caused immense problems, probably still costing them money today.

0
0
Silver badge

Re: Business Evolution

It would be nice to think that at some point, most of your cost are energy, and you don't need to spend much on developers anymore. However, I would bet that companies which make that choice soon find themselves ran over by the competition.

As a data point, it seems that Google uses about 5 millions MWh per year, which should cost on the order of 200 million dollars per year. Google has 50'000 employees, which are certainly more expensive than that.

0
0
Bronze badge

Re: Business Evolution (Google Employees)

$200mil / 50k = $4,000/year. So, depending on how many of those are H1B visa holders, I wouldn't bet the bank on those odds (kidding).

0
0
Gold badge

Simply running dockerd idle induces a 2 watt difference in average power

From the paper, that's 64kJ over 10 minutes rather than 63kJ.

1
0
JLV
Silver badge

bring it on!

The math that is...

"Simply running dockerd idle induces a 2 watt difference in average power".

This is the only quantitative info in this article. But it, crucially, it does not give us a the baseline : +2w out of 20, 200, 2000?

@Ken mucho txs for the clarification but it should have been in the original article. We can handle it.

As to uselessness attributed to academics by other commentards...

assuming those academics didn't use too much time coming up with those numbers, we now know more than before even if that comforts us that the inefficiency doesn't matter all that much. At the scale where really big data centers guzzle watts, that's a nice to know.

1
0

Re: bring it on!

From: https://arxiv.org/pdf/1705.01176.pdf

In this paper, we compared the energy consumption

of various workloads running within Docker-managed

containers and on “bare-metal” Linux. After almost

2 days and 20 hours of total time collecting power

measurements, we found that, in all cases, workloads

running in Docker have a measurable energy overhead.

Simply running dockerd idle induces a 2 watt

difference in average power, and thus an increase in

energy over time. However, the increase in energy

consumption may mostly be attributed to runtime

performance. In the case of Redis and WordPress,

the increase in energy can be attributed to increase in

runtime—thus the decrease in performance explains

the increase in total energy consumption.

0
0
Facepalm

NEWS FLASH: Doing more uses more energy

4
1

Next in line: virtual machines perform worse than equivalent bare metal

Scoop!!! Shock and horror!!! Who could have thought that adding abstraction layers could introduce... GASP!!! overhead!!!

This has to be investigated. And studied. And a herd of consultants will start shortly selling tools and processes to avoid this huge cost increase that all these fools using virtualization never realized before. Should we call them... ahem "server huggers"?

2
1

Academically but probably not practically interesting

These studies always have an interesting academic angle (although in this case the results seem a little obvious), however the details never take into consideration the bigger picture.

I'm unlikely, for example to be able to fully max out multiple bare-metal Linux servers to the same efficient degree I can running containers. Dedicating hardware resources creates fragmentation that results in the loss of any savings. Just think of how workload demand curves change over the course of a single day for any given application.

Then there are questions on poorly written code, or even more efficient code that works better as a microservice than monolithic application. What about patching and maintenance? What about backup (backing up a stateful bare metal server may consume more resources than never backing up a stateless Docker host).

As I said, interesting academically with little practical use.

5
0
Silver badge
Holmes
1
0

Academic doesn't always mean intelligent

I find it funny the amount of money wasted in Grants for Academics to carry out studies that not only state the obvious but are a waste of time and money.

Yes, having something running (i.e. a container) will require more energy - but you also need to look at the total cost of ownership along with ease of use, maintainability and management. Plus the number of hours extra over the year it would take to manage and execute the various pieces in the software value chain.

But, in the spirit of Diplomacy, maybe the Academics could design a solar-powered Docker container container ?

Thank you and good night !

0
0

Re: Academic doesn't always mean intelligent

Looking at the study I see it's pretty short and to the point, I doubt any grant money was needed.

1
0
Silver badge

Re: Academic doesn't always mean intelligent

"Looking at the study I see it's pretty short and to the point, I doubt any grant money was needed."

All they did was stick in a smart meter from British Gas and watched it for a day.

I could do that.

0
0
Anonymous Coward

This article seems like hooey.

But I have been struggling of late to see the benefits of Docker containers for big enterprise / production applications that are tied into many other things (Windows services are always a good example of a mish-mash)

0
0
Anonymous Coward

Wonder if they considered....

I haven't read the article (from the university) yet but it looks from what I mentioned that they may not have considered the following:

-The energy to produce the extra bare metal systems

-The energy to transport the extra bare metal systems to the warehouse

-The heat (different form of energy) in the warehouse where the bare metal systems were stored for a few weeks to months

-The transport cost of bringing them to a facility where they would be used

-The extra energy in cooling for the extra bare metal systems.

-The extra energy to create a larger server room to house the extra bare metal systems.

-The extra energy at end of life to remove all the extra bare metal systems and bring them to a recycling facility.

-The extra energy involved in recycling the extra e-waste at end of life.

-I'm sure that if I spend more than the 5 minutes I did writing this I could come up with a few more costs in energy.

This doesn't include the larger amount of e-waste create by have more servers that will be tossed away over time. Things are not as straight forward as they would seem but if you choose to look at the world through a pin hole it makes sense that things appear upside down (meaning counter-intuitive). It's funny 15 to 20 years ago penny pinchers were complaining that a system would sit idle most of the time so they were paying for power that did nothing.

Just my two cents on the topic

1
0
Anonymous Coward

Re: Wonder if they considered....

Things don't just magically appear at the front door when you ask for them... there are many things that are done before they get there.

0
0

Headline grabber

Surely a better comparison would be between docker and a full fat VM, and then docker on a VM. You'd imagine that paying the virtualisation energy price would only be done once.

0
0

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Forums

Biting the hand that feeds IT © 1998–2017