OpenStack is on the verge of greatness
For fellow Stackers, this statement is obviously opinionated. But for those new to or in the early stages of exploring OpenStack, let us give you an objective view, a teaser if you will as to why we feel this statement is true.
Some interesting developments are taking place in the community. These developments are focused on the ability to deploy and manage OpenStack. The quick summary here is that as these developments mature — they will provide the leverage needed to accelerate the adoption of OpenStack by orders of magnitude.
Today, OpenStack while having years of solid development under its belt is still lacking features in the areas of production deployment and management. OpenStack is known for its complexity and existing components provide little in terms of operational awareness of the infrastructure. OpenStack will need a minimal level of feature completeness in order for broad adoption to take hold in the marketplace. OpenStack deployment and management as a community initiative should ideally provide a long-term and stable platform upon which we can collectively build value.
Here's a quick summary of 5 of the top OpenStack feature requests from the latest Havana user survey:
- Simplified installation and configuration process
- Rolling migrations with N-1 compatibility
- Enhanced security
- Admin functionality
- High availability for OpenStack components
Clearly, the "asks" here are related to OpenStack deployment and management. What is generally not known is that the community has responded by addressing the comprehensive and complex task of deploying and managing OpenStack as elastic cloud infrastructure. In comes the OpenStack Deployment program.
The OpenStack Deployment (and Management) Program
Today, OpenStack provides very little visibility into the performance, capacity or service locality of its infrastructure. This lack of visibility severely limits cloud capacity planning and ongoing infrastructure management. Further, these areas limit confidence in the technology by those assigned to deploy the next generation of cloud services within their organization.
Ok, but why is this critical for the mainstream adoption of OpenStack?
Our vision for OpenStack Deployment and Management is to deliver a functional, community led framework that can provision OpenStack at scale while providing a foundation for feature growth in the areas of planning, deployment and ongoing operations.
For those of us in the community, time is of the essence. Our joint participation in the OpenStack Deployment program opens up a path for us to answer the needs related to the planning, deployment and on-going operations of OpenStack.
So let’s take a look at the 3 areas of focus for the OpenStack Deployment Program.
1. Infrastructure Modeling
A Red Hat initiated a project named Tuskar was recently merged into the OpenStack Deployment program and provides the ability to model the infrastructure of an OpenStack deployment. In short, this means that information about infrastructure locality, resource node (compute, object storage, block storage, etc) awareness and the relationships between each can now be described and used by OpenStack.
Today, we are often effectively blind to what's happening with our infrastructure. Tuskar allows us to map and describe the underlying OpenStack resource infrastructure. With Tuskar we can now encapsulate, visualize, track and manage service levels and resource states. Because Tuskar provides the logical map of what we’re running, we can now ask for metrics and other data representing the state of the hardware supporting our cloud. Before we can actively manage the infrastructure, we need to have situational awareness. Tuskar's infrastructure modeling is the foundation that gives us this awareness.
HP has led the charge to tackle the OpenStack deployment problem space. OpenStack installation and deployment is a complex process, typically beyond the scope of even large, well-staffed organizations. Communal knowledge related to elastic cloud deployment best practices is at an early stage and is still squarely in the intellectual hands of Amazon, Google and Facebook. Most OpenStack deployment issues arise from traditional, enterprise approaches being applied to OpenStack in an unnatural fashion. Our goal here should be to make OpenStack very easy to deploy while adhering to and embracing elastic cloud infrastructure best practices.
3. Operator Dashboard
The OpenStack Deployment program now includes foundational support for an OpenStack infrastructure dashboard. The operator specific dashboard is intended to allow views into OpenStack infrastructure performance, capacity, and core service status. Prior to OpenStack Deployment becoming an official program, we lacked a community led framework providing capacity planning support, instrumentation or operational and deployment awareness of OpenStack infrastructure and core services.
OpenStack's continued maturity is dependent upon making the life of those deploying and managing clouds easier. The OpenStack cloud operator persona has largely been ignored as we've been writing code for ourselves since Bexar. As an open source community we are competing against large and well established organizations providing tools in a proprietary but uniformly integrated software stack. The "competition" innately understands that the cloud deployment and cloud operator personas are customers. We as an open source community didn't. The thinking has changed and in that process not only did we learn who the customer was, but how to best provide essential services.
Infrastructure modeling, easy deployment and cloud operator dashboards - oh my! Is this really OpenStack? Yes!
Red Hat is working with HP, Intel, Solinea and others to solidify the work being done today. We are excited about making OpenStack easier to use for worldwide adoption and specifically as a downstream benefit for our customers. If you are in Hong Kong for the OpenStack Summit, join us to see some of this work in action!