Ensuring High Availability for SAP in an Oracle Linux Environment

Anyone investing in an SAP ERP solution running on Oracle Linux is clearly serious about ensuring that their critical line of business systems run smoothly, predictably, and efficiently. It should go without saying that those operations will be consistently accessible and run without unexpected interruption. Saying nothing, though, would be a mistake, for there’s nothing magical about Oracle Linux that inherently guarantees high availability (HA) for an SAP landscape. Without proper planning and an infrastructure configured for HA, that big investment is vulnerable.

So, what are your options? Rely on the data replication systems built into SAP? Configure Oracle Real Application Clusters (RAC) for HA? Deploy something analogous to the Pacemaker HA solutions found in SUSE and Red Hat Linux distributions? There are third-party HA solutions, but SAP solutions are notoriously complicated and unless you deploy a third-party HA solution that is built to support SAP in an Oracle Linux environment you may not secure the HA guarantees you seek.

SAP Solutions for HA

Let’s start with the data replication systems built into SAP. If you’re using SAP HANA, you might consider SAP HANA System Replication (HSR). HSR makes a copy of the SAP system—including the services and the in-memory database—and replicates the data to a secondary system. Should the main system fail, a system administrator can manually bring the secondary system and database online and redirect system traffic to that infrastructure.

One problem with HSR is that it is an incomplete solution. It replicates a critical subset of the SAP services and data that are active in the landscape, but it only replicates that subset. To ensure access to all SAP services and data—which is important if you discover that you need to rely on your backup infrastructure for more than a very short period—you’ll need a separate replication solution (in addition to HSR) that can ensure that the SAP services and data sets that HSR ignores are properly replicated to secondary infrastructure. For example, if you are running other databases in parallel to HANA, such as MaxDB or an Oracle database, HSR will not ensure the high availability of these systems.

One other problem, as noted above, is that HSR does not provide any orchestration services to execute an automatic failover from a failing active node to a passive node (the replication target). To gain automated failover capabilities, you would have to rely on a solution external to the SAP landscape itself.

The issues of partial replication and lack of automated failover orchestration components can leave your enterprise vulnerable to unexpected downtime, and no organization investing in an SAP solution will find that tolerable. Lack of native failover orchestration features could delay bringing your backup infrastructure online as quickly as your HA needs demand. The lack of a complete replication of system and database data may compromise system performance and/or integrity if you find yourself relying on your backup infrastructure for more than a few minutes. If you need access to the services and databases that were not replicated under HSR, you’re out of luck unless you deployed an additional service to replicate those aspects of your SAP landscape.

Oracle RAC

What about using Oracle Real Application Clusters (RAC) to ensure HA for your SAP landscape? It may be possible. Oracle RAC enables you to distribute a single database across multiple servers in order to maximize availability and enable horizontal scalability. But Oracle RAC is designed to support Oracle applications and databases, not applications and databases from third parties such SAP or Microsoft (which is important to keep in mind if your organization is thinking about running SAP on Oracle Linux or a Microsoft SQL Server database). Furthermore, Oracle RAC is very expensive and complicated to manage.

Other Options for Ensuring the HA of an SAP Landscape

What about a tool like Pacemaker HA? It’s not a native component of an Oracle Linux distribution (as it is in SUSE and Red Hat Linux distributions), so you’ll have to download and install it separately. More problematic, though, is that Pacemaker’s manual approach to deployment makes it extremely error prone—which is not at all welcome, given the complexities of the SAP landscape itself. System administrators that are already highly skilled in Pacemaker may encounter no challenges here, but for the average system administrator, Pacemaker is cumbersome and difficult to deploy. It also does not validate any of the configuration settings, which increases the risk of errors that you may not even know are present until something goes awry at a critical moment.

As for third-party HA solutions? Third-party solutions designed to ensure HA in an Oracle Linux environment can provide the full range of services you’ll need to ensure the HA of a complex SAP landscape. The key to success, though, is to find solutions that have been specifically designed to provide HA support for an SAP landscape—and not all tools can claim that.

For example, a third-party solution built with an SAP landscape in mind will provide a mechanism for replicating all the SAP services and data from primary to secondary infrastructure—in separate locations to eliminate the vulnerabilities associated with a single point of failure. That’s foundational. An HA solution built with an SAP landscape in mind will go further. It will possess the intelligence to know how to orchestrate the failover of different SAP components so that if a failover is required the standby infrastructure comes online in the proper sequence.

It’s worth noting that some third-party HA solutions for Oracle Linux have even more tools in their kit than just a well-orchestrated failover. Some solutions provide monitoring features that are specifically tuned for SAP landscapes. They can proactively monitor not just the heartbeat of the servers supporting various components of the SAP landscape but they also monitor key operating system, disk, network, and application services. Besides monitoring these services, the HA solution can identify and, in many cases, proactively address events that could lead to problems later on—problems that might compromise performance of the SAP system or even lead to a condition triggering a failover to the secondary infrastructure. By recognizing and remedying these conditions before they reach a critical point, these tools can improve the performance and the availability of an SAP solution without even having to resort to a rapid and well-orchestrated failover.

One final note about third party HA solutions: Many organizations configure their SAP solutions in a heterogeneous manner. Some parts will run on Oracle Linux; others will run on Microsoft Windows Server. Solutions such as Oracle RAC and Pacemaker may offer specific advantages for those components running on Oracle Linux but they will not ensure the HA of those components running on Windows Server. Among the third-party HA options are product suites that run in both Linux and Windows environments. If your organization is—or may eventually be—running your SAP landscape with multiple underlying operating systems, this becomes a critical point to consider. HA for SAP is not truly HA unless all the components of your SAP landscape are configured for HA. The weak links and single points of vulnerability are the elements that can bring even the most resilient infrastructure to a screeching and unwelcome halt.