PUMA SE is one of the world’s largest manufacturers and retailers of footwear, apparel and accessories for athletes in a variety of sports. Its mantra is Forever Faster, an imperative to be “the fastest sports brand in the world.”
PUMA’s commitment to speed extends to how it interacts with an increasingly digital world. The business depends on fast applications used by personnel around the globe along with suppliers, distributors, dealers, company stores and online retailers such as Amazon. The applications are powered by in-depth data about products, parts and prices contained in the main production database, which must operate at lightning speed. As Herbert Wirkner, Technical Architect and Senior Database Administrator at PUMA, says, “If no data is going out, no products can be sold.”
Previously running on IBM® Db2® software with High Availability Disaster Recovery (HADR) in two server clusters, the database supports production applications for product development, sourcing, online commerce and other functions. Because application users frequently request new digital services and features such as custom interfaces and APIs, PUMA’s IT managers made a strategic decision. Rather than customizing core applications directly, they would develop new functions as cloud-based microservices in Red Hat® OpenShift® containers.
“Microservices are part of going Forever Faster,” says Wirkner. “Using them, we can add functionality to the applications much faster than before.”
Still, deploying many new microservices will increase the database workload. This led to a question around how PUMA could ensure that its database environment would have the power and scalability to process the rising transaction load.
Faster upgrades
Deploying the IBM Db2 pureScale Feature was a major upgrade that took just 6 months from design to going live from design to going live
Huge capacity
Load testing showed that the new system can support 300%–400% more users than before
Workloads at scale
If microservices increase database workloads, the system can be scaled out to 128 servers
Another query involved the business-critical nature of the database. It needed to be available 24x7 worldwide, yet the system had to be taken offline for routine maintenance to the software and hardware. Since downtime is unacceptable, the PUMA team sought a more highly available environment.
To brainstorm solutions, Wirkner turned to ARS Computer und Consulting, a long-term IBM Business Partner and an expert in software engineering and development. He chose ARS for its longstanding support of PUMA’s database technology and its close ties to IBM Db2 experts from the IBM Canada Software Lab in Toronto.
Potential solutions for more performance included partitioning the database across servers, but this didn’t suit the database format. Upgrading the HADR server clusters was also explored, but this wouldn’t provide scale-out capabilities. The right solution, the team agreed, was enhancing the database with the IBM Db2 pureScale® Feature, a cluster solution designed for scalability and continuous availability.
“From PUMA’s perspective, Db2 is an excellent source for their microservices to access and processes companywide data, but they needed more performance,” says Roland Schock, Distinguished Engineer at ARS and IBM Gold Consultant. “They wanted to scale out to multiple servers to process the workload and pureScale was the answer.”
PUMA, ARS and IBM personnel began architecting the system in late 2019. Their task entailed deploying new physical server clusters and high-speed networking components while enhancing the database software.
As year-end approached, the developers faced a watershed moment. They had not yet gotten the pureScale feature operational and the budget for the project was set to expire in December. Should they wait, or go ahead and spend the budget on the assumption they could resolve any glitches?
“We trusted in IBM and bought the licenses for two pureScale clusters, one for production and one for testing,” says Wirkner. “It was the right decision because now they’re really running great.”
Each of the two pureScale clusters consists of four physical servers running SUSE Linux® Enterprise Server across four server nodes. Four “member” nodes run database workloads and two nodes manage caching and coordination. The system enables PUMA’s database to be accessed by concurrent instances of Db2 running across the different members, all using shared virtual storage and cooperating as a single system.
The architecture delivers significant processing power that can scale out to as many as 128 servers. The performance gains are nearly linear—IBM reports that overall throughput almost doubles as the number of members doubles.
To help ensure availability in case of disaster, the server clusters are split across two nearby data center locations. If one data center fails, the other is ready to take over. And by integrating several advanced hardware and software technologies, the system supports the strictest requirements for fault tolerance and can process database requests even under extreme circumstances.
Wirkner and the PUMA team are quite pleased with the project’s results. Time to value was fast. From design to going live, the project took just six months, a time that was prolonged when the pandemic sidelined offices and staff.
“I think six months is good for such a big project, especially as we had to implement new hardware, network devices and other things,” says Wirkner. “Together with ARS and the Db2 experts from IBM’s Toronto software lab, we had the perfect team to bring the solution to life.”
Equally positive are the performance numbers. PUMA used test scripts in Apache JMeter to simulate an increasing number of users accessing the database through their applications. The load tests found that the four-server pureScale cluster supported four to five times more users than before. With scalability up to 128 servers, PUMA should be set for whatever the digital future brings.
Availability, too, has improved from several pureScale features. It allows server upgrades without taking the system offline, and a failed server causes another member to take on its workload.
Similarly, routine software maintenance can be done with online fix-packs without impacting the system. “That is a really great feature, I no longer have to ask for downtime of a production cluster,” says Wirkner. “I can apply the fix-packs on the fly without interrupting the business.”
More database speed, scalability and availability to power microservices — PUMA gained these benefits from pureScale. Now, the stage is set to deliver faster digital innovation.
Headquartered in Herzogenaurach, Germany, PUMA (link resides outside of ibm.com) designs, develops, markets and sells sports-inspired footwear, apparel and accessories. Targeting athletes in sports such as running and training, soccer, golf, basketball and motorsports, PUMA operates in 120 countries and employs more than 13,000 people. The company reported sales of EUR 5.2 billion and gross profits of EUR 2.5 billion in 2020.
Based in Munich, Germany, ARS (link resides outside of ibm.com) specializes in software engineering and development services. An IBM Business Partner at the Gold level and an IBM Champion, ARS was founded in 1992 and employs around 70 people. In 2018, ARS became part of TIMETOACT GROUP, which employs more than 700 people at 16 locations in Germany, Austria and Switzerland.
Legal
© Copyright IBM Corporation 2022. IBM Corporation, Hybrid Cloud, New Orchard Road, Armonk, NY 10504
Produced in the United States of America, January 2022.
IBM, the IBM logo, ibm.com, Db2, and pureScale are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the web at “Copyright and trademark information” at ibm.com/legal/copyright-trademark.
The registered trademark Linux® is used pursuant to a sublicense from the Linux Foundation, the exclusive licensee of Linus Torvalds, owner of the mark on a world¬wide basis.
This document is current as of the initial date of publication and may be changed by IBM at any time. Not all offerings are available in every country in which IBM operates.
The performance data and client examples cited are presented for illustrative purposes only. Actual performance results may vary depending on specific configurations and operating conditions. THE INFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS” WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING WITHOUT ANY WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND ANY WARRANTY OR CONDITION OF NON-INFRINGEMENT. IBM products are warranted according to the terms and conditions of the agreements under which they are provided.