Endress+Hauser maximizes availability and resiliency with Linux on System z

To read a German version of this case study, click here.

Headquartered in Reinach, Switzerland, Endress+Hauser is a specialist in measurement technology for process engineering. The company develops and maintains instrumentation and automation solutions for industrial processes. Employing 8,400 people in 89 companies across 42 countries, Endress+Hauser generates annual sales of approximately €1.1 billion.

SAP ERP applications are central to Endress+Hauser’s operations, supporting the company’s core business processes worldwide. Approximately 5,400 users depend on the company’s 100 SAP systems, which are used to manage the entire production-line for measurement equipment, as well as the sales and distribution departments. Originally, these ran on 14 distributed servers, mirrored to a secondary data center, which resulted in an overly complex disaster recovery strategy.

“Quite simply, the failure of our old solution would translate into a lot of work for our IT team,” explains Aribert Starnell, Division Manager, Production Computer Center at Endress+Hauser InfoServe, the internal IT service provider of the group. “We would need to transport servers to our backup locations and then carry out some complex configuration before migrating the data and booting up again. Equally, it wasn't clear that we would have full vendor support for a distributed recovery option.”

Need for easy manageability

Maintaining high availability is a major priority for Endress+Hauser, especially as the company offers some of its clients web access to certain information related to their orders. As a result, maintenance windows are restricted to just two Sunday half-days per year. The complexity of the prior system with distributed servers was making it difficult to carry out necessary updates within the strict designated period. The aim was to find an alternative solution that would help the company’s relatively small IT team manage the systems efficiently and effectively.

Starnell comments: “Many other companies employ much larger IT teams to manage a similar number of users. But we were confident that we could find a simpler solution that would enable us to stick to our tight maintenance schedules whilst preserving maximum uptime.”

Winning cost-benefit analysis

Endress+Hauser was a long-time IBM mainframe customer, running five IBM DB2® databases on an IBM System z® machine. Based on its experience of the disaster recovery capabilities of the mainframe—and, in particular, the IBM support around these capabilities—Endress+Hauser decided to look at the possibility of moving the SAP environment to this platform also.

“Working with IBM, we conducted a detailed cost-benefit analysis to compare a Linux® on System z solution to alternatives such as IBM Power® servers or x86-architecture machines,” says Starnell. “We were very encouraged by the results. The analysis showed that moving to SUSE Linux Enterprise on System z was comparable in price to continuing with our existing platform. We were eager to harness the cost-effectiveness and stability of Linux combined with the processing power of a mainframe. Most importantly, we wanted to have all critical systems in the same environment to make disaster recovery simpler, faster and more reliable.”

As part of its selection process, Endress+Hauser looked at the total cost of ownership for the various disaster recovery options. A critical element in the calculations was the manpower costs, which were found to be considerably lower for the mainframe—in part because the SAP environment would slot into the existing disaster recovery arrangements. Equally, the sophisticated workload management capabilities of the mainframe give Endress+Hauser very granular control over which services are prioritized in the event of a disaster.

The company migrated its SAP application servers to two IBM System z10® Enterprise Class mainframes running z/OS® and z/VM®. It also activated 80 IBM Integrated Facility for Linux (IFL) engines to provide a highly available environment for new SUSE Linux Enterprise Server workloads.

The two IBM System z10 servers are both located in the company’s primary data center in Weil, Germany, and are clustered using IBM Parallel Sysplex. These servers are protected by a diesel generator, enabling them to continue running in the event of a power cut. All data is synchronously mirrored to a secondary data center, approximately 60 km away in Weil. In the event of a disaster, IBM will deliver an additional System z server to this secondary location that can be connected immediately to the backup hard disks and booted up again in a short amount of time.

The backup z10™ server is much smaller in terms of CPU and memory (to keep costs low), and does not offer enough performance to handle the full workload. The ability to control the allocation of resources to the most critical services means that Endress+Hauser can ensure effective recovery without the high cost of maintaining a one-for-one equivalent of its production environment.

“Our disaster recovery strategy is so much simpler and more reliable now,” comments Starnell. “Whereas previously there was a daunting amount of manual configuration effort involved in booting up the systems again, now it could not be simpler. IBM will handle delivery of the mainframe, and we can simply connect it to our backup storage and be up and running again in no time!”

He adds, “There are many options for disaster recovery, and we could have done this in other ways. However, we felt that the support and the proven disaster recovery capabilities of the mainframe made it the best option.”

In addition to Parallel Sysplex, Endress+Hauser uses IBM DB2 Data Sharing across the two System z10 servers in the primary data center. As well as providing extremely high availability, Data Sharing simplifies routine maintenance, enabling the company to move all DB2 workload to one z10, patch the other server, and then return it to the cluster—all without any service interruption.

DB2 for z/OS offers further benefits for the disaster recovery of SAP environments. One notable feature is the federated backup capability, based on the Parallel Sysplex timestamp. This makes it possible to recover all elements in an SAP landscape to exactly the same point in time, eliminating the need for the complex and time-consuming logical corrections that may otherwise be required to reach a consistent state across logically connected but disjointed SAP system boundaries.

Boosting user satisfaction

With the migration to SUSE Linux Enterprise Server carried out successfully in stages, Endress+Hauser soon began to discover the benefits. Using IBM System z, the IT staff at Endress+Hauser can move workload between machines as required, enabling them to roll out updates and patches without downtime.

“We had no issues with the migration—in fact, none of our users even noticed the change-over,” says Starnell. “What they did notice, however, was an improvement in response times as the performance of the Linux operating system on System z is so good.”

“One of the biggest advantages for our IT teams is the ability to switch between the IBM System z machines easily and quickly. This helps us ensure 100 percent uptime outside our very limited maintenance windows, maintaining the high availability that it is so important to our business.”

Making the most of resources

By leveraging virtualization, Endress+Hauser is able to provision new Linux environments quickly and rapidly in response to sudden demand. The solution provides the necessary flexibility to scale capacity to meet the needs of different applications automatically, with no need for manual intervention.

Starnell explains: “The combination of SUSE Linux Enterprise Server for System z and IBM z/VM enables the creation of new Linux virtual instances in a matter of minutes, making it easy to deploy new SAP systems whenever the business requires. The superb scalability of the solution gives us the confidence that we can easily accommodate future growth.”

The easy manageability of the system makes it possible for Endress+Hauser to maintain its relatively low staffing level despite increasing number of users.

Starnell concludes: “Other companies that visit us are often amazed that our team is so small—but the simple and intuitive user management tools for System z make it possible for just 1.5 full-time employees to administer the entire Linux landscape, whilst two run z/OS and one monitors z/VM. We have found z/VM so stable that only one person is required for emergency cases. This has convinced us that SUSE Linux Enterprise Server for System z is the right choice for our mission critical SAP applications, in terms of both ease of use and cost-efficiency.”