During the last few weeks we have met several companies interested in Infrastructure as a Service, and they all loved what Openstack Nova promises . The idea of having a truly open platform that has an ecosystem of companies around is very attractive. Openstack participants are companies with consolidated reputation and recognized by professionals. Everything seems perfect … Everything? Not quite…During the conversation there is a question that is constantly raised:
How can I deploy Openstack in my data center?
That’s the point when we start to look to each other… because an Openstack Nova deployment is (well, it was) far from trivial. You can spend days going through its documentation and most likely will not get beyond a single server installation . It’s fine for a quick overview, but far from what an Infrastructure as a Service should be.
Recently some interesting tools appeared allowing automation of the deployment of the nodes. That’s great, but I think they forget something that is key to the deployment of Openstack Nova: this is not only a server farm, it is a platform on which services are connected to each other at a higher level than the operating system, making necessary something or someone that ‘orchestrates’ the deployment.
We realized that most people go through three basic stages in the deployment of a standard solution with Openstack Nova: First contact, Proof of Concept and Pilot, plus a final stage, Production deployment. The number of companies performing a first contact is very high, but for obvious reasons those who reach production stage are the least. The potential market for companies that will ultimately put into production (which will therefore become clients) may be less than 1% of the First Contact installations, less than 10% of those that deployed a Proof of Concept, and less than 50% of those who tried a pilot with their customers. It’s impossible to size a startup to attend directly to potential customers, especially in the first two phases. Therefore, if we wanted to reach a very high audience and then scale globally we should be able to define deployment architectures that could meet the demands of each stage with minimal support from our side, hence unattended and automated.
And of course … it should work. A Systems Administrator must find a level of difficulty similar to installing a VMware ESXi or Citrix XenServer.Neither more nor less.
And that’s how we started to build the Stackops Nova Openstack Distro. A distribution based on Ubuntu 10.04 LTS Server that contains everything needed to deploy a Openstack Nova solution. It includes strictly what is needed to run Openstack Nova. The distribution can be configured so that it can take different roles of architecture: Controller, Volume, Network and Compute.
But a bare-metal distribution does not solve the problem of configuring a deployment: you need something else. And that’s why we developed the Stackops Smart Installer. The Smart Installer is an agent embedded in the distro that talks to a web application hosted by us and supports the administrator in the deployment of any of the reference architectures shown above. Do you want to deploy a virtual machine for a quick test? Ok, no problem. Do you want an architecture with multiple nodes? The Smart Installer keeps those global parameters of the architecture and inject them as needed. Global and local to components parameters hell is over . The assistant is responsible for keeping track of them.
But the best part is that the Distro and the Smart Installer are free. You can download the ISO and test it right now on a virtual machine,or perform a complex attended deployment if you have a server farm. Of course, if you need customization, contact us