edit Copy Return Up-pointing chevronRight-pointing chevronfacebooklinkedintwitter

Not All Heroes Wear Capes…

I found myself thinking about the amount of time that we spend on the web nowadays. Almost all our jobs involve spending hours on the internet, corresponding with remote team members or customers, creating marketing campaigns, building websites and apps… And in our free time we indulge in long social media breaks, FaceTime conversations or stream movies on Netflix with our phones.

For all of this to happen and our devices to connect and function on the internet we need some special computers called servers and we need power to make them work, too. Servers let us enjoy the internet from the comfort of our screens but also can run the code to build a website, store data and let us do other amazing things! For example they allow us to run hydro simulations to make sure that dams run properly in an under-the-sea-level country like the Netherlands, but also calculate chances of Tsunamis in Japan and they find their use in the med tech industry and biology labs as well, for example in the case of protein folding or HIV and cancer medicine research.

Usually servers are collected in buildings called data centers, which can range from a small room with a few machines to humongous farms with over 150k square meters of surface. There are approximately 9 million of these data centers across the globe, out of which almost 400 are hyper scale, in total taking up to almost 836 square kilometers of space, an area bigger than New York City.

For us to be able to be connected and computing at all times on both sides of the equator, these data centers have their servers working full time, every day of the year, thus using about 10% of all the power utilized by the United States’ Government. Well, that’s also because if the internet were to go down even just for a handful of minutes, companies would lose billions of dollars (an estimate calculates $5,600 of loss per minute of IT downtime)!

Imagine now the amount of energy used by the people working for the data centers to commute every day, the power needed to make the servers work and to keep the data centers up and running and sum it all up. Huge, right? Yet, the problem is another.

You are probably familiar with the heat that your laptop generates and radiates on your legs when you’re overworking in bed on your out-of-office day. Now picture the heat that all the servers in a data center can produce. In most data center aisles, the servers cause the temperature to be around 27 to 46 celsius degrees. The ASHRAE (American Society of Heating, Refrigerating and Air Conditioning Engineers) recommends a temperature in between 18 and 27 celsius for servers to operate in. The difference between the actual and needed temperature can then be quite dramatic and needs to be taken care of to avoid putting the servers’ operability at stake. Well, for a computer to work you need a ventilation system and for a data center to function you need air conditioning. Which means that those aforementioned 836 square kilometers need to be cooled down constantly, causing further pollution at an impressive degree. In fact, the energy to cool down the data centers is even more than the needed amount to make the servers function!

Data centers now utilize over 3% of the global electricity supply, not to mention that their greenhouse gas emissions account for approximately 2% worldwide, placing them in the same carbon footprint area as the airline industry.

Our internet consumption is constantly on the rise, having become some sort of pleasurable addiction for the average millennial and a crucial tool for scientific progress, both in terms of research and simulations, but this brings to an issue. If some companies are now trying to use renewable sources of energy to bring power to their servers and allow us to connect to the Internet of Things in a less guilty way, what about all the energy that is dissipated trying to cool those servers down? Some options are starting to emerge in the market yet all with some considerable downsides. Placing servers under the sea, like Microsoft did with Project Natick, might reveal itself to be problematic in case of need of maintenance. Not just that, it would not reutilize the generated heat but just dissipate it in sea water. We at Nerdalize are offering a very clear and convenient solution, because we don’t believe in zero sum games.

Why build a data center if you can put servers in people’s homes and re-utilize the energy used for the computations to heat up their water? That’s why at Nerdalize we have built a heating device based on computer servers. Instead of heating up data centers, we heat up water to 55 celsius degrees. One of our server-heaters can supply 80% of the needed heating for a household of four. This lets us save up 2 tons of CO2 per home per year and the benefits are not just for the environment! The Dutch citizens hosting our servers in their homes get cheaper energy, saving approximately €200 per year on heating. The lack of need for physical space to contain our servers is one of the factors impacting our cost structure, allowing us to offer a cloud service, for simulating water levels or folding proteins, at a price that is 40% lower than our competitors’ with no compromise on performance. Not all heroes wear capes, but they sure do all care about the planet!

About our sustainability
Elena Galli
posted this October 8

September Changelog

Portal

Creating Kubernetes clusters is now much easier: we’ve made it possible to do so directly on the portal and there’s clear information about the size and cost per cluster.

We’ve improved the cluster information with status labels and better resource charts.

Playground cluster

We’ve fixed an issue, now making it possible to create deployments using kubectl.

Fast video transcoding with FFmpeg in the cloud

Video processing and especially transcoding are quite CPU-intensive tasks. That’s why doing it in the cloud is great. Not only can you easily use a large number of CPUs to make your conversions go faster and to run many side-by-side, you’ll also have your computer free to do other work. And by running it on the Nerdalize cloud you’ll be able to do so affordably, while heating someone’s shower and saving CO2 at the same time.

FFmpeg logo

FFmpeg is a great tool for video processing, but like many applications it’s usually quite a hassle to run it in the cloud. To make scaling to the cloud easy we’ve made it super-straightforward to run FFmpeg on Nerdalize. Using our FFmpeg Docker image and Nerd, our easy-to-use CLI, you can transcode any video file from your computer in the cloud using just two commands:

$ nerd job run \
	--input=~/my-video-files:/input \
	--output=ffmpeg-output:/output \
	--memory=3 --vcpu=2 \
	nerdalize/ffmpeg -i input.mp4 output.avi
$ nerd dataset download ffmpeg-output ~/my-transcoded-video-files
Submitted job: 'ffmpeg-run'
To see whats happening, use: 'nerd job list'

Downloading (Step 1/2): 972.80 KB / 972.80 KB [=======] 100.00% 0s
Unarchiving (Step 2/2): 972.80 KB / 972.80 KB [=======] 100.00% 0s
Downloaded 1 dataset

Of course it’s not just limited to converting MP4 files to a AVI. You can use the full FFmpeg featureset like creating optimized videos at different resolutions and bitrates, various other compression options for video and audio, creating thumbnails, etc. In fact, you can use FFmpeg just like you’re used to by passing any of the normal FFmpeg arguments. We had some fun with it in our example job.

A screenshot of our example image

Try FFmpeg yourself for free

We’ve created a short guide to get started with running your FFmpeg jobs on the Nerdalize cloud. By registering, you’ll get free cores so you can process your own videos right away, or try it out with our example video.

About our cloudsoftware
Alexander Weiss
posted this September 19

August Changelog

Docs

There’s a handy copy button next to all example commands now.

All example commands now come with additional information about the arguments and options used as well as a button to show their output.

Most arguments in the example commands are now editable.

Portal

We’ve significantly improved the loading speed of the portal.

Nerd

We’ve improved the speed of logging in and listing clusters.

Network to the Home with the speed of light!

Building a datacenterless cloud is quite the challenge. To illustrate, Microsoft researched heating homes with servers in their Datafurnace Project but (among other things) ran into the issue of the low average bandwidth in the US.

At Nerdalize we’ve always believed we can offer datacenter-grade cloud, including low latencies, without actually building a datacenter. We’re now at the point where we actually have 10 Gb/s connections to the homes our servers heat. Nerdy enough to want to learn more? Let me explain our network setup to you!

10 Gb/s and sub-millisecond latency

Luckily around 1/3rd of the homes in Netherlands have Fiber-to-the-home (FttH). Nerdalize has built a network, reusing much of this existing FttH infrastructure. By deploying our own network equipment and uplinks to the neighbourhood, we were able to set up a datacenter-grade network over existing fiber, with 10 Gb/s bandwidth to each household and at a latency between 0.15ms and 0.20ms.

Our network

And since all homes are equipped with two physically separated fibers, homeowners’ and cloud customers’ traffic can also be completely separated.

A datacenter-grade network allows our customers to run a wide variety of processes on our servers. We want to offer the same or better performance as any cloud provider, but in a more sustainable and affordable fashion!

The best part? We don’t charge you network costs. Ingress and egress are free under a fair use policy. Try it for yourself! Create a free trial account and check out how fast your application can run outside of a datacenter.

About our cloud
Maaike Stoops
posted this July 26

July Changelog

Docs

New guides will help you get started with Kubernetes on Nerdalize.

Portal

We’ve removed the need to manually edit your kubeconfig file by providing kubectl config commands that you can just copy and run.

Nerd

Jobs now default to using 1 vCPU and 1 GB memory.

We’ve improved some of the feedback Nerd provides, especially when there’s an error.

Start heating homes with your Kubernetes!

Last month we gave an exciting talk at KubeCon about how we use Kubernetes to heat the Netherlands. We’re happy to share that you can now also run your Kubernetes clusters at Nerdalize!

Our Kubernetes offering is now in public beta for you to use. Create your next Kubernetes cluster in the Nerdalize portal and start heating homes with your pods.

These affordable clusters come with some great features:

Kubernetes support out-of-the-box

Heat homes with your private Kubernetes clusters.

Freely scale between 1 and 80 vCPUs

Flexibility to choose a cluster that fits your scale at the press of a button.

More affordable than the rest

With € 0.026 per vCPU hour we’re cheaper than other providers. Check our full pricing here.

Multi-cloud dashboard included

We offer a multi-cloud dashboard, so you can use other providers or your on-premise cluster as well.

If you register for an account you’ll get free access to 8 vCPU to try out in our Playground cluster.

Do you need more capacity? Add your payment details to get your own multi-cloud dashboard and start using Kubernetes at Nerdalize or other cloud providers or try our easy-to-use Nerd tool.

Multi-cloud Dashboard
Spin up Kubernetes at Nerdalize and others with our multi-cloud dashboard

About our cloudsoftware
Maaike Stoops
posted this June 11

Lyon/Delft, April 25th, 2018. iExec and Nerdalize have signed a partnership agreement towards providing sustainable cloud solutions to blockchain applications. Nerdalize therefore officially sells its cloud resources through the the iExec decentralized cloud computing marketplace.

Partnerships Towards Green and Energy-Efficient Computing

iExec continues to gather computing resource providers that will act as the first workers on its blockchain-based decentralized cloud marketplace. iExec’s prior collaboration strategy has been to partner with not only distributed, but also sustainable and green computing providers. Earlier in the year, iExec teamed up two other companies, Stimergy and Cloud&Heat, that share a similar vision of providing energy-efficient computing resources.

It is increasingly becoming common knowledge how colossal the energy consumption needed to support traditional data centers is. The running and cooling of data centers have a combined electricity consumption often greater than that of entire nations, while C02 emissions from powering servers exceeds the whole airline industry.

Nerdalize: Sustainable and Affordable Computations at Scale

The vision of the Netherlands-based startup ‘Nerdalize’ is to eventually move away from traditional and centralized data centers. The company’s innovation is to place servers directly in residential homes and recycle the energy that dissipates from these servers into heating water. This process results in a reduced heating bill for homeowners, while removing the cost of setting up a traditional data center.

Moving servers to households saves 2 tonnes of CO2 per household per year, as well as the overhead cost of building a data center.

 Maaike Stoops, Business Developer at Nerdalize

Nerdalize owns servers called ‘CloudBoxes’ that are installed in households all across the Netherlands. Dockerized computations are distributed to multiple CloudBoxes to run jobs for an impressive 40% of the cost of centralized cloud providers.

Powering the First Decentralized Marketplace for Computing Resources

iExec’s V2 release planned for May 29 presents a new paradigm for cloud computing: a global and open market where computing power is traded like a commodity. The marketplace offers an easy-to-use interface through which applications and users in need of computing power can meet providers able to answer their demand at the best rate possible.

Nerdalize has joined the initial group of private workers in the iExec marketplace. By combining efforts, our two companies are not just leaving a positive mark on the environment, but on society as well by making cloud computing an affordable commodity for everyone.

Wassim Bendella, Business Developer at iExec

By not having to invest in building and maintaining expensive data centers, iExec and Nerdalize can provide more affordable cloud computing offers. This in turn allows developers to innovate more easily in the compute-intensive fields of artificial intelligence, IoT, scientific research or big data, therefore radically enhancing our existence in unprecedented ways.

About iExec

iExec is developing the first blockchain-based decentralized cloud marketplace. This marketplace will allow developers and users in need of computing power to meet cloud providers willing to answer their demand. iExec The network supports the most compute-intensive decentralized applications in the fields of AI, big data, fintech, cryptography or 3D rendering.

About Nerdalize

Nerdalize is building a different cloud. Instead of constructing huge data centers, the company is distributing its servers over homes. Homeowners use the residual heat for hot showers and to warm their house, without needing to build new infrastructures. Nerdalize enables users to scale up computations without wasting CO2 or money.

Our new and improved Nerd CLI is here

We have some exciting news to share with you. We just released a new and completely overhauled version of our Nerd CLI!

After introducing our beta a while back we’ve received great feedback from many of our users. Taking all this input into account, both minor and major issues, we’ve revamped our Nerd CLI.

A couple of the major improvements:

  • Easy Windows installer — Get running even faster.
  • Give your jobs and datasets a name — Stop struggling with difficult identifiers.
  • One command to rule them all — Upload a dataset, start a job and get it to run in just one step.

You can find the full release notes for the second release candidate on our Github.

We’re again looking for users to try Nerd for free. When you’ve registered, download Nerd and login to start your first job by running:

  1. Login

    $ nerd login
    
  2. Run your first job

    $ nerd job run --name=co2-calc nerdalize/co2-calculator
    

If you have any feedback or questions, keep sending them over!

About our software
Maaike Stoops
posted this March 19

Running Python scripts in the cloud — Super easy!

At Nerdalize, we’re trying to make computing as easy and straightforward as possible. Last month we released a Python Docker image that allows you to run your scripts without creating a custom image. You can just upload your script, requirements.txt and data files and run it right away. We’ve got a Python 2 and a Python 3 version, both of which are built on the official Python Docker image.

Don’t have an account for the testing period? You can use the following promo code for early access: try-python.

Try our example script: The CO2 calculator

We also created a nifty little CO2 calculator that you can use to try out our Python image. Just follow the quickstart to calculate how much CO2 is saved by Nerdalize each year. It’ll show you:

1) Total CO2 savings in kilograms and travel distances (by car, train & plane). For example, for just a few households, the reduction of CO2 emissions is equal to almost 4 car trips 🚗 around the world 🌎. The CO2 calculator will give you the exact number.

2) A chart showing how many airplane trips you could make to various destinations.

Read more Right-facing arrow
About our cloudsoftware
Liesanne Wieleman
posted this January 25