Procter & Gamble moves some mission-critical manufacturing to the cloud with Azure Ultra Disk Storage

For nearly two centuries, Procter & Gamble (P&G) has developed innovative products that today help improve the lives of people all over the world. With an employee presence in 70 countries, and its products sold in more than 180 nations worldwide, P&G has put a lot of work into ensuring that consumers have the opportunity to obtain its industry-leading brands. 

“Imagine you’re in one of our 60-plus plants across the globe, and today the raw materials you had delivered changed slightly,” says Tim Elftman, Operations Lead at Procter & Gamble. “You need to know what recipe changed and what adjustments you might need to make.” To address these concerns, and anything else regarding how it sources, creates, and manufactures its products, P&G uses a dedicated application. “That’s our mission-critical manufacturing app, used for everything from putting ingredients together to running them through the manufacturing process,” says Elftman.

This manufacturing solution has traditionally run in a high-speed, on-premises environment that enables large user volumes to routinely access and utilize even larger stores of data. When P&G decided to move much of its infrastructure to the cloud, including the nine different environments and 160 virtual machines across 12 databases that support the solution, the search was on to find a suitable cloud-based environment. “We looked at all the major cloud providers,” says Matt Fleck, PLM Global Platform Manager at Procter & Gamble. “Because of the critical nature of these systems, we needed the transition to be as seamless as possible. Microsoft understood that we needed to drive performance and stability, and it was willing to work with us toward that goal.”

Easing the move to the cloud 

Due to the complexity and performance requirements of moving this manufacturing solution to the cloud, this would not be a simple “lift and shift” migration. Working with Microsoft, P&G went through multiple virtual machine (VM) configurations, including different storage disk types. After implementing each configuration, P&G ran user-centric performance and load tests designed to simulate heavy usage. “We typically take a single VM, simulate 40 concurrent users, and then run that same VM at load,” says Fleck. “That way, we can actually see the impact to performance as we ramp things up. We then also run that same test over an extended period of time to see if there is any degradation in performance.”

But with the constraints of sending data between P&G facilities and the cloud, the available VM configurations had a tough time matching the throughput speeds of an on-premises deployment. “We took those results back to one of our daily meetings with Microsoft, and our timing couldn’t have been better,” says Fleck. “They had just made a new, faster storage option available—Azure Ultra Disk Storage. We were one of the first to test it out.”

Azure Ultra Disk Storage offers high throughput, high IOPS (input/output operations per second), and consistent low-latency storage for Azure infrastructure as a service (IaaS) VMs, creating top-of-the-line performance at the same availability levels as any other Azure disk offerings. Ultra Disk Storage can also dynamically change disk performance as workloads increase, without the need to restart VMs. “Once we implemented Ultra Disk Storage, we had our eureka moment. We knew that moving to the cloud would be a challenging; however, it was less daunting once we saw the throughput those disks could give us.” 

Getting up to speed

P&G is quick to point out that, while Ultra Disk Storage played a crucial role in moving the application to the cloud, it had more to do after that important first step. “All of a sudden, the performance we’d been hoping for was there,” says Fleck. “We were meeting, or in some cases exceeding the IOPS of our on-premises solution, but next we had to make sure that the rest of our newly implemented cloud infrastructure could keep pace.”

P&G ran benchmarks and tests against baseline data from the company’s on-premises infrastructure. Together with Microsoft, it began fine-tuning details at the operating system level and the VM level, including changing the instance type it ran. Networking speeds were accelerated, and some Microsoft E-Series VMs were swapped out for F-Series VMs in scenarios where both CPU and memory-intensive workloads are commonplace. It was an iterative process, with both companies working to refine P&G’s cloud infrastructure until, much like Ultra Disk Storage, the platform could meet or exceed an on-premises implementation.

To ensure everything went smoothly for this mission-critical migration, Microsoft worked hand-in-glove with teams from Red Hat, a leading provider of enterprise open-source solutions, including high-performing Linux. “Ninety percent of our systems interacting with the manufacturing solution are Linux-based,” says Silesh Sankaran, Cloud Architect for ITS Engineering at Procter & Gamble. “The rest are Windows Server–based. That’s been the case for a very long time, so part of the migration process was enabling Azure to manage that heterogeneous environment at speed.” 

With the migration and fine-tuning process now predominantly complete, P&G says the operation of its mission-critical manufacturing solution in the cloud is going well. “With Ultra Disk Storage, we saved time and effort in the midst of what was an incredibly complex and crucial move,” says Erin Simons, Director and Project Manager for Non-SAP Cloud Migration at Procter & Gamble. “The capacity and speed of Ultra Disk Storage helped ensure a smooth transition to the cloud.” 

More benefits on the way

Since the migration, P&G has also moved a handful of other workloads with high IOPS requirements to Ultra Disk Storage. The intuitive IOPS/throughput adjustments provide flexibility even for the company’s non-production environments. “We’ve gained a lot of agility, architecturally speaking,” says Fleck. “We’re using cloud-based tools to gain better visibility into our processes, enabling the identification of any potential slow-downs and driving both efficiency and value.”

For Fleck, the journey of fine-tuning and balance continues. “The key thing for me,” recalls Fleck, “was discovering that Ultra Disk Storage, running at the lowest possible throughput, matched the highest throughput we could achieve before. As we learn, as we grow, I think we’ll see this turn into a cost-saving endeavor as well, whether that means adjusting how we take advantage of the flexibility inherent to Ultra Disk Storage, our VMs, or how we can now share and leverage our workloads across virtual environments.”

Find out more about Procter & Gamble on Twitter, Instagram, YouTube, and LinkedIn.