Nimbula 'cloud operating system' spans data centers
Ex-Amazon man transcends geography
Nimbula – the build-your-own-cloud outfit founded by Amazon's former vice president of engineering – has announced a new release of its Director platform, saying it will allow businesses to run a unified "infrastructure cloud" across geographically separate data centers.
In short, this means that those using such a cloud can log into a single console to tap computing resources served up from disparate physical locations. Nimbula says that Director is the first "cloud OS" to span geographies in this way.
Reza Malekzadeh, Nimbula vice president of marketing, tells The Register that Director version 1.5 will be available in September.
Installed on bare-metal servers, Nimbula Director is designed to mimic Amazon's EC2 infrastructure cloud – but inside the firewall. From your own data centers, it offers up readily scalable computing resources such as processing power and storage.
The platform uses the Xen and KVM hypervisors, and it offers a REST (representational state transfer) API that also maps to Amazon's APIs. This lets you move workloads between Amazon and a cloud based on Nimbula. Director is free if used with server clusters of fewer than 40 cores. For larger clusters, you'll pay an annual subscription that provides maintenance and support services.
Director 1.5 will also include a new storage setup similar to Amazon's Elastic Block Storage (EBS), a means of creating storage volumes that can be attached to virtual server instances, and it will allow administrators to offer end users prepackaged operating systems – complete with drivers and management software – that can then be deployed onto virtual servers.
Nimbula was founded by Chris Pinkham, the former VP of engineering for Amazon Web Services, the Amazon arm that runs EC2. He envisions Nimbula Director as something that will overhaul the way internal IT resources are handled.
"We want everything to be automated from the first time the infrastructure is turned on," he told The Register before the platform launched. "We hope that this will be a significant contribution, changing the way that traditional infrastructure is managed and thought of. We think of it as infrastructure-as-a-service out of the box." ®
His Generic Holiness...
There is no such thing as 'generic clustering', so this is just hype.
Only established things like content-serving can make use of any sort of clustering solution 'out of the box', everything else must be tailor made.
This is a Holy Grail matter, ie. not really solvable and it will make big CompSci news if it ever is.
Why do I keep saying 'cluster' instead of 'cloud' ? Because I'm a computer scientist and I call things what they are, I don't care what some coked up exec calls them.
I'm no clould expert but....
How is this different than the new availability zones feature in OpenStack or even EC2? Or is it that Nimbula does away with the zone concept all together?