Availability and Recovery

IT infrastructure and SAP architecture

Many years ago, I spoke with the IT manager of the Austrian retail group Spar on the occasion of the renovation of its data center at the Spar headquarters in Salzburg. In addition to new servers and storage for SAP ERP/ECC 6.0, he rented a failover data center. Both locations were connected with two fiber optic cable strands - under the stipulation: At no point could the cable strands be buried closer than 200 meters to each other in the open field. The request is logical: an excavator is not uncommon. An excavator rolling uncontrolled 200 meters should be almost impossible.

SPOF

Theoretically, there should be no single point of failure in any IT system. However, due to the complexity of hybrid ERP landscapes, this requirement hardly seems feasible. Existing SAP customer Lufthansa was probably not in a position to include the networking of the outsourced wide area network in its own risk analysis. Perhaps there is also a service level agreement according to which such a dredging accident cannot occur - all theory is gray.

Cloud First

Regardless of the IT construction sites at Deutsche Bahn and Lufthansa, every existing SAP customer should realize that cloud computing is much more than lift and shift. The redundancy of the hyperscalers' data centers is largely limited to their internal, global networking. The data path to the user is already a new IT chapter. With cloud first, many challenges are shifted, but not solved. There is a danger that cloud computing will cause many problems to slip out of the company's own sphere of influence. What was a daily task in the company's own data center operation seems to become invisible in the cloud.

High Availability

High availability is an expensive commodity and must be used carefully. The head of IT at the Spar retail group knows full well that the failure of the SAP system would very quickly lead to empty shelves in the stores and long queues forming at the scanner checkouts. Two data centers with widely separated fiber optic connections are therefore not a luxury, but a business necessity. The on-prem CIO can see the high availability for himself every day. As a cloud administrator, the CIO must hope for the reliable work of others.

Disaster Recovery

If the worst comes to the worst, a back-up plan must be available. With all resources in the company's own data center, there is the possibility of a quick emergency plan. With cloud computing, the CIO first has to look for cause, effect and responsible parties in the cloud. It seems that some things run better with cloud computing, but the administrative effort itself is a lot higher. Why did it take so long to find the source of the problem at Lufthansa IT? A broken fiber optic cable is not rocket science.