CoreOS: A new way to building servers

Old No Comments on CoreOS: A new way to building servers 36

CoreOS is a Linux distribution with fresh take on clustering, virtualizing and deployment.

Clustering is the process of seamlessly maintaing a group of servers to operate as a single entity with relative ease. With CoreOS, its as easy as creating a cloud-config file. A cloud-config file will contain the basic information of your server cluster as well as contain an unique URL to identify clusters. The URL is generated at https://discovery.etcd.io/new and this unique identity makes joining a server to a cluster ridiculously easy. Once you add your unique URL on the cloud-config file to a new CoreOS server, it will instantly join the cluster and be visible to all others within it.

One of the unique aspects of it is that it contains NO package manager. The base operating system merely a means to build, hold and run containers as well as manage the communication between servers. It runs all neccessary services in isolated containers managed by docker.

Docker is a version of operating system-level virtualization which relies on micro-images of various distributions to form a base to build containers from. Docker allows interfacing with containers by command line to build and run applications needed for your server then opening up only the necessary ports to allow traffic. This creates a significant security improvement since this allows fewer avenues of approach for unauthorized access.

Some of the caveats you should be aware of is that it doesn’t support setup of disk quotas or I/O rate limiting. Also, it doesnt allow partition checkpointing or live migration nor does it allow a comlete root isolation.

Fleet is perhaps the crowning jewel for CoreOS. It works with docker to ensure that your container(s) are running, even in the event of maintainance failure. This is done by utilizing etcd across all members of the cluster. This allows any member to run containers seamlessly even in the case of a catastrophic multi-point failure. This can allow even ensure 100% uptime if configured completely.

It’s worth giving it a try. With so many features, it gives system administrators ease when it comes to rapid deployment and management

    Sources:

 

Author

David Hile

I build both Linux and Windows servers for fun.

Related Articles

Leave a comment

Back to Top