Lets get the liability out of the way so I don't get in trouble with my employer. This blog is completely my own opinion.
With HP's Helion OpenStack and Carrier Grade OpenStack solutions shared at Mobile World Congress this year, one could think customized and proprietary solutions are the way of the future for service providers.
Far from the truth, both community and proprietary solutions need to contribute to the same software community which brings us to my April post. What sets different OpenStack distributions apart and how does this look to both business owners and technologists.
Long post short, we will need to use OpenStack provider enhancements outside the community while features continues to mature. This is where HP Helion OpenStack Carrier Grade comes into play.
If we think of all service provider vendors today and their appliance based business, we can see a majority make their revenue off hardware.
When a SP CTO looks at this model and tries to map their business plans to a virtualized or OpenStack world, they may have mixed feelings. An open platform with minimal vendor locking is key to an OpenStack adoption. On another hand they could see the hardware cost reduced (theoretically) and an increase in licensing (this is where those vendors will recoup their lost HW profit).
From a technology perspective, OpenStack is an event horizon for IT departments, helping to bring their teams to the next frontier.
Now lets think of our first adventure into the land of virtualizing our carrier grade appliances and solutions. If we take mobility as our first example, we need to first find out what these vendors offer and their NFV requirements. Once we have a general understanding (SR-IOV, NUMA, Huge Pages, DPDK) of all the features we then look at our OpenStack options. Which of these features are available in OpenStack and which need to be developed by an OpenStack provider.
At this point, our path to OpenStack adoption veers towards vendor lock-in for the purpose of enhancing features beyond current community efforts. Ideally we would want OpenStack providers to upstream their customizations but the community may not always accept these changes.
Lets take this one step further and imagine we have our custom OpenStack environment with all the features needed. Our licensing model may cost more but we have what we need to do business and guarantee things work.
Now say within the first year of our cloud being operational, the community stabilizes all the original features we needed. What now? This is where our OpenStack provider like HP needs to provide a seamless upgrade path for the cloud to reduce our customized software and redeploy community base OpenStack onto existing hardware.
This model in my opinion is the best strategy to accelerate time to market as well as reducing CAPEX related to normal hardware/software refresh.