Internet Infrastructure for Everyone 63
just_another_sean sends in a story at Wired about a group of engineers trying to build a new server operating system that will make it easier to deploy a multitude of technologies for people and companies that aren't tech giants.
"The project is based on Google’s ChromeOS, the new-age laptop operating system that automatically updates itself every few weeks, but unlike ChromeOS, it can run more than just your personal machine. It can run every web service you ever visit, no matter how big. And it will let the companies that run those services evolve their online operations much more quickly — and cheaply — than they can with traditional server software. 'We’ve borrowed a lot of concepts from the browser world,' Polvi explains, 'and applied them to servers.' You can think of CoreOS as a new substrate for the internet. Web giants such as Google and Amazon and big Wall Street financial outfits, including the NASDAQ stock exchange, have built similar server operating systems for their own use, but with CoreOS — an open source software project — Polvi’s startup is creating something anyone can use. 'We’re building Google’s infrastructure for everyone else,' he says. In doing so, Polvi and his team hope this OS can more rapidly fill the security holes that plague our computer servers, while speeding the evolution of the software applications that run atop them."
Sounds like a bunch of bullshit to me (Score:5, Interesting)
Can anyone decipher exactly what it is he's promising? Another layer in the OSI model that tries to reinvent the Java wheel and run everything natively?
Re:Sounds like a bunch of bullshit to me (Score:4, Interesting)
It almost sounds like their trying to tie server services into the cloud... probably not actual data storage, but the services and functions themselves. Kind of like how if you go install chrome on a new computer, it can port over all of your settings and stuff, or how if you setup a new android device it will automatically load up your apps and contacts, etc.. In this case, I think the idea is to make it so that the hardware is less important, and easier to replace without having to go through the normal motions of reloading from backups or doing some kind of barebones restore. Instead, you just swap out or install whatever new hardware you need to, and "long in" (or whatever the process may be) to get your new server node online and sync'd with the rest of your network without much hassle.
There's a lot of blanks that need to be filled here, like actual data store. I imagine that would still be done in-house with central storage. The basic idea, as I understand it, is actually really cool.