CoreOS: the cloud system of tomorrow?

It is the open source project to follow. A Linux extreme who suffered a weight loss program, and can be deployed on thousands of instances in minutes.

 CoreOS, what is it?

CoreOS The project was born with the ambition to recreate what the open source web giants like Google are doing internally to manage their server farms. Google, Facebook or Twitter manage thousands, tens of thousands of machines but without automation tools, deploy a new feature or perform a simple security update can take considerable work to system teams. CoreOS offers both Linux architecture optimized for server and tools needed to manage large or very large clusters.

 Who supports the project?
google datacenterRecréer open source that web giants are internally to manage their server farms: this is the ambition of CoreOS. © Google

Like any legend in Silicon Valley who respects himself, CoreOS was born in a Palo Alto garage. At the head of the project, Alex Polvi. After creating Cloudkick, a startup he sold to the giant Rackspace Cloud, this former Mozilla launched the project with CoreOS Michael Marineau, a former Google and Brandon Philips, a Linux developer. One of the first to support the project was Lew Moorman, president of Rackspace. Since CoreOS received financial support from Andreessen Horowitz and Sequoia Capital, the two most prestigious Silicon Valley investment fund.

 What does it include?

The basic brick CoreOS is a minimalist Linux, optimized for performance servers. CoreOS designers are Chrome OS parties to develop a Linux stripped to the extreme, a Linux that starts very quickly (one instance is launched in just 2 seconds!), And that consumes a minimum of RAM, 161 MB depending its authors. Besides the OS, multiple components are used to create and manage very large clusters of treatment.

What does it include?

The basic brick CoreOS is a minimalist Linux, optimized for performance servers. CoreOS designers are Chrome OS parties to develop a Linux stripped to the extreme, a Linux that starts very quickly (one instance is launched in just 2 seconds!), And that consumes a minimum of RAM, 161 MB depending its authors. Besides the OS, multiple components are used to create and manage very large clusters of treatment.
fleet scheduling coreosOutre the OS, multiple components CoreOS to manage very large clusters of treatment. © CoreOS

 What is the use of each brick?

Besides the OS itself, the key components of CoreOS are all from recognized open source projects. Thus, Docker is the system used for container CoreOS. It is he who will place the application to be deployed in a separate container which will then be deployed on a large scale over thousands of Linux instances.
architecture CoreOS 2Plusieurs application containers can be placed on an instance. © CoreOS

Cluster Configuration and management are centralized through a number of DCE bricks: a daemon that runs on each node to ensure consistency of the cluster, systemd to manage starting and stopping services and finally for the Fleet Management global cluster.

The architecture proposed by CoreOS allows distributed applications very quickly on the cluster. It provides redundancy and fault tolerance, and dynamically reallocates resources.

 What are the application areas?
architecture proposed by CoreOS coreosL’architecture to quickly distribute applications on the cluster. It provides redundancy and fault tolerance, and dynamically reallocates resources. © CoreOS

Cloud service providers and hosting providers are the first interested in the work of CoreOS but a wide variety of applications can be “containerized” in Docker and thus deployed in a cluster CoreOS. The Index Docker full of containers ready to be downloaded: databases, PHP engine, Python, Ruby, Java servers, applications like SugarCRM or OpenERP, load balancer, etc. The great strength of CoreOS is to allocate just enough these containers in the cluster based on available resources and the load required by applications.

 With which cloud platforms does the solution support?

If the sites looking for performance at any price will deploy CoreOS on their own machines in a standard approach “bare metal”, it is easy to deploy CoreOS on all major existing public cloud services. Amazon EC2, Google Compute Engine, Rackspace and Brightbox services are cited as CoreOS can run. It is also possible to deploy CoreOS on a private cloud architecture, either on VMware, OpenStack and Eucalyptus servers, but also Libvirt, QEMU and Vagrant.

 Who uses it today?

If Alex Polvi said that many companies are testing CoreOS, including Fortune 500 companies, no great player has not officially reported using the solution.
Eric Barroca nuxeoEric Barroca, CEO of Nuxeo. Open source editor has chosen to CoreOS to support its cloud offering. © Nuxeo

In France, Eric Barroca, CEO of Nuxeo open source editor, has chosen to CoreOS for future cloud offering. “We are preparing an offer of PaaS around our platform and we are interested in CoreOS this purpose we Docker already use to manage our bodies and we looked for a solution to rotate the Docker containers as well as possible, with a minimum of overhead. ”

He added: “CoreOS done it all very well, but what is very interesting with this solution are the tools that go around: the OS is designed to run in the cluster, it is very light and full coordination among the nodes cluster, including via DCE is extremely interesting. ” Seduced by the solution, the French involved in the development of the platform through the GoMeta project, a reverse-proxy that distributes the load on the CoreOS cluster.

 How it fits into an existing infrastructure?

For Eric Barroca, logic by container CoreOS questions the current approach, in which the virtual machine is created by applying “CoreOS is an innovative approach, this is clearly the next generation platform that will succeed virtualization. It keeps the benefits of virtualization, while adding since the density can run multiple containers on a single instance CoreOS. “