Architecture must fit the landscape
If you want to optimize your SAP architecture, the first question you need to ask yourself is: What do you want to achieve with it? Only when there is an answer to this question can your IT landscape be optimized.
For example, do you primarily want to reduce high administrative expenses and the costs of your IT operations or would you rather rely on a strategic platform for the future?
Companies should consider this carefully, as the strategy of how they approach the matter and what measures they should implement will depend on this answer. They can either carry out this process on their own or seek advice from specialized consulting firms.
In any case, it is important to base decisions on facts rather than on assumptions and perceptions. An important prerequisite is therefore to analyze the IT infrastructure:
How many SAP solutions are running on the platform, and how many should there be in the future? How available and fail-safe are the systems, and what is planned? Are changes to the infrastructure already underway? Is an in-memory conversion pending?
The data collected in this way can be used to identify weak points, bottlenecks or requirements from specialist departments. Furthermore, the selection of possible target platforms can be narrowed down from the outset.
Mirror, mirror on the wall, ...
... which is the best platform in the country? Many companies ask themselves this question, for example, when their hardware lease is coming to an end or has been written off.
This is a good thing, because the platform is of central importance for the optimal architecture of SAP solutions. In the classic SAP ERP area, there are several to choose from: Microsoft, Linux, HP-UX or IBM AIX as well as OS/400 and Solaris.
The reasons for switching from one platform to that of another provider can be manifold: for example, because it allows the IT landscape to be consolidated or because another platform is considered more future-proof than the one currently in use.
At the moment, the keywords in the SAP environment are Hana and Linux. The extent to which SAP solutions can be scaled or virtualized on a platform and the costs involved also need to be examined.
Years ago, many companies switched from Unix to Linux because it is very stable and compatible with other systems. In addition, the hardware costs for CPU performance, for example, are lower than for other platforms.
In addition, many procedures can be retained and existing knowledge can be used during such a changeover, which minimizes the amount of training required.
Availability is a tricky subject. The very definition of what is meant by high availability can sometimes cause companies problems. Does high-availability refer to the guaranteed operation of a system or does it include the ability to absorb the possible failure of an entire data center?
For security reasons, the physical distance to a second data center must be correspondingly large. On the other hand, it must not exceed a certain distance because otherwise it is difficult to keep the data synchronized.
If there is one data center in Europe and one in the USA, they are not 100% synchronized, as the data cannot be mirrored in real time at this distance without affecting the application.
Avoid head monopolies
Less is sometimes more: when it comes to architecture, it's not just hard facts such as the hardware price or availability that count. What use is a cheaper platform if it does not optimally support the processes or the employees do not have the necessary training?
If companies want to switch to a different platform, the question also arises as to whether the employees have the necessary specialist knowledge and, if not, what training they need to complete to ensure that the new platform is well accepted and operated.
A platform change should always be accompanied by change management processes. These should not only relate to the technical migration, but also prepare employees for the changes.
This also helps to avoid head monopolies, where knowledge is concentrated in just one employee. If this employee is absent or leaves the company, their expertise is also lost.
It can therefore be advantageous to choose a technically simpler platform that several employees are familiar with, rather than a sophisticated and highly available premium version that only one employee can master.
However, manufacturers' offers are generally aimed at technical performance. This can lead to the hardware being oversized or "over-engineered".
The motto for companies is therefore: the platform must remain operable. Less is sometimes more. Especially as the costs for additional training can amount to up to 15 percent of the technology costs.
Professional consulting firms prepare the results of the as-is analysis, their recommendations and a list of the necessary expenses in the form of a meaningful study.
The advantages and disadvantages of possible platforms are also listed. Based on this, the next steps can be decided on the basis of facts.
When migrating, it is advisable to start with a smaller system, as a test so to speak, followed by the remaining systems - possibly grouped together.
Performance data should be collected after two months in order to measure success. If you wait longer, further adjustments - IT landscapes are subject to constant change - could make a comparable measurement impossible.
Examples of successful migrations to Linux
The Generali Group decided to modernize its SAP infrastructure. Around 100 servers and databases, 24 system lines for 14 countries and 100 terabytes of data were to be transferred to a new target platform.
The existing infrastructure was to be brought up to the latest technical standard with the help of a service provider and, at the same time, dependency on a single hardware supplier was to be reduced in order to ensure the future security of the investment.
After a workshop, roadmap, risk matrix, analysis, cost estimate and a proof of concept, the systems were successfully migrated to Linux with sustainable profitability, two months earlier than originally planned.
The proof of concept in particular was an important success factor for Generali because it gave the Group the certainty that it had chosen the right strategy.
In less than 20 months, Munich Re also transferred its server architecture, including 84 SAP systems, to a Linux platform. Some of the data was also converted to Unicode during the migration.
Change management posed a particular challenge, as the company was introducing the largest SAP application globally at the same time as the migration.
In addition, infrastructure operations were transferred to an offshoring model. Thanks to special cut-over plans, up to six SAP systems could be migrated simultaneously in individual weeks.
All cost targets were achieved without affecting the business or the availability of the applications. Munich Re succeeded in reducing the procurement costs for SAP servers to a fifth of the original investment volume.
Since the migration, applications and databases have been running on servers with a standardized process architecture. Standardization has also significantly reduced operating costs.
Demand will increase
In addition to aspects such as scalability and availability, holistic architecture consulting also takes organizational and financial factors into account. If a consultancy firm does not also sell hardware, the customer can be sure of independent advice.
When migrating to a new platform, the focus should no longer just be on technical performance, but also on strategic components such as future-proofing.
It is also advisable to review and possibly adapt the processes. An approach based on ITIL is ideal for this. In the SAP environment, Hana and the possibilities of the cloud will ensure that the demand for architecture consulting increases over the next few years. Additional expertise will be required.
Architectural consulting based on strategy and foresight will nevertheless guide you calmly to your new target architecture with a coherent concept.
Nevertheless, the order of the day remains: the decisive factor should not be what is technically feasible, but always what suits the landscape. See the Norwegian opera house.
