This month we (discreetly) interviewed a cloud migration pioneer who helped to establish an enterprise-grade cloud migration factory within a multi-national company of more than 10,000 employees. To be specific, we’re talking about migrating existing, on premises workloads onto AWS.
We asked him about: 1) business drivers behind their decision; 2) how his group approached cloud; 3) key obstacles and how they were overcome; and, 4) what the cloud migration factory means for the team and the company going forward.
(Clouds): Thank you for talking to Clouds. We reached out because we knew readers will be intrigued with what your team has accomplished. Could you start by telling readers about the cloud migration factory project and the business drivers? What was it about the cloud that made it so compelling?
Answer: We have been on the journey for the last few years, which has given us a bit of an edge. We learned how to tailor our approach to adopting cloud. We view cloud as a platform not just a capability. While this notion didn’t sink in immediately, over time it shaped our thinking on how the cloud could drive maximum value for our business stakeholder.
Our cloud migration factory has resulted in savings for IT and has enabled our organization to focus on things that matter most to our business.
We had three primary considerations: 1) 30-40% of our workloads are non-production. They aren’t required to be on all of the time; 2) we are required to maintain system state for legal/compliance purposes for 7 to 15 years; and 3) data needs to be readily available for e-discovery.
We took a look at our dev/test farm of 200 servers and calculated that each server was costing us about $7k/year to run/support. By migrating that workload to the cloud and using it on demand, we have reduced costs by over 45%.
Then we reviewed how we addressed data and system preservation, which is a key requirement for companies like ours. We have data and systems that are not transacted upon, yet do require ongoing support. Our legacy model for supporting infrastructure doesn’t differentiate between transacted versus non-transacted systems. So we were incurring high internal costs by maintaining a 100TB, 5 file-server system. The cloud allowed us to reduce our annual storage costs by 85%, a considerable savings over the next 10 years.
Our previous approaches to e-discovery were time-consuming, expensive and risk-prone. So we moved the data to object storage, enabled ELK services for indexing the data and gave our e-discovery teams a self-service view of searching/tagging the data that was more relevant to their analysis. This was further enhanced by enabling additional cloud object storage which we could simply convert into a web service that could be accessed by our processing agency. This not only saved us money, but enhanced agility.
(Clouds): What were the key considerations for the project, in terms of available clouds, internal skills and approach, and external resources? Did your team have any internal advantages that made it easier or harder?
Answer: We started with small win projects. Our approach was to show some wins and gain some understanding as to what it took to run/operate cloud services. We took baby steps, starting with IaaS services, then over time we moved more toward PaaS.
Our decision regarding which public cloud to adopt was based on who provided the most compliance/security, and we identified what other companies like ours were doing. I think we all have been learning from each other.
As far as internal skill were concerned, part of our challenge was how we initially defined the cloud. There was lot of focus around infrastructure, so we initially assumed it was yet another data center with advanced hypervisor capabilities, and if we could simply use on demand services we could save money.
After investing over 3 years into cloud projects we realized it required different DNA and thinking to properly run and operate. So we shifted our focus to establish a Bi-MODAL organization chartered to lead disruptive technologies initiatives. That is when we starting making progress both from a speed of adoption standpoint and in gaining some key wins. This was also the time we started understanding what it would take and the type of resources we would need to run and operate in the cloud. This lead us to develop a dedicated team made up of people who understood infrastructure and came from a software development background.
I don’t think we had any specific advantage as our challenges were no different than many other organizations. What it came down to were individuals driving each initiative forward. Individuals who were not afraid to fail, continued to identify use-cases, developed proof-of-value solutions and looked for quick wins.
(Clouds): What kinds of obstacles appeared as you planned and then executed your program? Were there any surprises (good and bad)? What about any “been there, done that” recommendations for your peers contemplating cloud migration for existing premises workloads.
Answer: CULTURE, CULTURE & CULTURE – You are trying to influence two decades-old operating behaviors, something to which enterprise are not accustomed. If your CIO/Senior leadership is not committed then don’t waste your time. You will need their support. Moving to the cloud requires new thinking, for everything from running a file server to running a J2EE application. I also recommend that you consider a Bi-Modal organization.
(Clouds): How do you see the cloud migration factory impacting business operations going forward? Will the payoff be as substantial as originally envisioned? What kinds of impacts do you see it having on your team’s ability to leverage the cloud for existing workloads?
Answer: As long as the organization sticks to the plan there can be a huge impact in terms of reduced costs, and enhanced agility.