BMC's four-year slog unleashes the benefits of hybrid cloud
The hybrid cloud that BMC had set out to build used a combination of on-premise infrastructure, both physical as well as virtual, while also leveraging public cloud capabilities from Amazon Web Services (AWS).
The project was headed up by BMC 's senior director of enterprise cloud services, Sandeep Kulkarni, and he explained that his team was able to maintain a flat budget by ensuring the architectural blueprint of the cloud infrastructure consisted only of standardised compute, storage, and network hardware.
"This allowed us to get away from an expensive maintenance agreement. Cloud provides a seamless fail-over, so honestly as long as I have spare blade, chassis, and network equipment, I don't need any maintenance of any sort. I can keep the spare in each of my major cloud providing sites and get away from the maintenance, which to me is just an insurance," said Kulkarni.
This is an important and seldom-mentioned attribute of cloud as the going rate for hardware maintenance agreements as a proportion of the hardware capital cost is around 20 percent according to Forrester analyst Boris Evelson, so it's quite a saving on the annual operating budget.
Of course, overcoming budgetary constraints isn't the only challenge in a migration project of this scale, with the team facing a number of technical and operational challenges throughout the course of the project.
The technical obstacles will be familiar to many IT departments that work in large organisations -- dealing with a vast library of legacy hardware and applications, more commonly referred to as 'brownfield'. In BMC's case, they had a lot of older HP, IBM, and Oracle physical hardware that needed to be virtualised and moved to the cloud.
Kulkarni says coming up with the right brownfield conversion strategy required a near-perfect understanding of all the complexities and nuances of the existing environment.
"Getting to know what you have and knowing what out of it is utilised is very important while building an internal cloud. Taking a service view of your IT is the first step towards a good cloud implementation. Once you know your business services, you can then map them into underlying technology services which are then fulfilled by infrastructure in the cloud."
The skillsets needed to evolve as well. The internal team at the time consisted of traditional Windows and UNIX network teams, and while BMC had the requisite network, compute and storage resources, domain expertise was highlighted as an area of deficiency.
"We identified the skills we had, identified the skills gap and trained the individuals to become true cloud engineers. We also added datacentre expertise to this group."
Operationally, processes based around the older physical infrastructure had to be redefined to the new automated cloud offering.
No pain, no gain
So was the four-year long slog worth it and did the hybrid cloud model deliver the gains the business originally set out to achieve?
"Absolutely," Kulkarni said. "Prior to our internal cloud, we were provisioning everything manually. This took days for our customers, also everything else in the system like configuration management database (CMDB), monitoring, decommissioning was also manual."
"The servers, once provisioned, would end up remaining on the network forever causing huge server sprawl. Our [storage] reclamation rate prior to our cloud was 4 percent and today after deploying cloud we are above 85 percent. This makes our infrastructure highly efficient.
"Our internal cloud provision times moved from days to minutes. And everything from CMDB to monitoring is all automated now. As a result, it freed up our internal teams by significantly reducing the time spent on managing the hardware. If we talk about software-as-a-service (SaaS), our tier 1 application availability has gone up multiple folds once we moved to the external cloud."
BMC today uses cloud in all of its capacity, almost 40 percent of its tier 1 apps are hosted on a SaaS cloud.
"We also manage our private cloud for product development, build and testing. The internal cloud started as IaaS, however we have matured the offering to not just operating system provisioning but also application stack provisioning," he said.
"We use external cloud in tandem with internal cloud for any scalability-type testing where the demand is not constant but fluctuates heavily. "
Kulkarni said that after the hybrid cloud was successfully deployed, there was a real shift in mindset amongst the team.
"With cloud services we moved away from asking our customer what server they want, to what services they want. That became the focus."