OpenStack (http://www.openstack.org/) truly has the imagination of a lot of the IT world right now. It has the power to be an incredible tool that changes how modern IT business is performed. It is actually a collection of tools that can run on a single or multiple systems and often have at least some overlapping functionality.
- OpenStack Compute (code-name Nova) – core project since Austin release
- OpenStack Networking – core project since Folsom release
- OpenStack Object Storage (code-name Swift) – core project since Austin release
- OpenStack Block Storage (code-name Cinder) – core project since Folsom release
- OpenStack Identity (code-name Keystone) – core project since Essex release
- OpenStack Image Service (code-name Glance) – core project since Bexar release
- OpenStack Dashboard (code-name Horizon) – core project since Essex release
Projects Incubated in the Grizzly release:
- Metering (Ceilometer) – https://launchpad.net/ceilometer
- Basic Cloud Orchestration & Service Definition (Heat) – http://wiki.openstack.org/Heat 
These tools allow you to see multiple types of hypervisors and storage systems through a single pane of glass. It is already a very powerful tool. The inclusion of the concept of metering and Orchestration in the latest release point in the direction of a solution, but they are (at least today) intended to be gateways into OpenStack more than complete solutions for those problems.
And that’s the problem that Enterprises will soon be facing. Even with Red Hat working on their supported version of OpenStack whose community is called RDO (http://openstack.redhat.com/Main_Page), OpenStack doesn’t provide all of the abilities that are needed to truly manage an enterprise installation and achieve a true ROI. So how do you do that?
A company has to be truly aware of what’s going on in their datacenters to really make money using a hybrid cloud approach at the level OpenStack can do it. It’s a matter of looking at a price sheet to see how much a given system would cost on Amazon for a length of time. But how much does it cost to add that same machine to one of your internal hypervisors? A company would have to be truly aware of heating and colling costs, and the changes made by increased load on a given hypervisor system to be able to accurately charge back cost to its various clients.
Once a company is aware at that level, there are a few ways to go. The one I’m currently very impressed with is from Red Hat. It’s the 2.0 version of their CloudForms product (http://www.redhat.com/solutions/cloud-computing/red-hat-cloud/private-clouds-cloudforms.html). When Cloudforms 1.0 came out I didn’t really see the goal or the point. However, Red Hat purchased a company called ManageIQ last year, and set their team on integrating their product in and working on the now-released 2.0 product.
It’s pretty nice for several reasons:
- It’s delivered easily as a virtual appliance (a first for Red Hat)
- It gives you the single pane of glass to automate and meter all of your virtualization environments
- It uses no hacky access methods. No database hacks. It goes through the front door and manages everything properly.
- It’s extremely easy to build out complex logic-driven events within your infrastructure
Laying CloudForms on top of OpenStack and your virtualization solutions gives enterprise users an incredibly powerful management platform for managing a truly hybrid cloud through a very few “panes of glass”.