Search on this Website

Wednesday, May 23, 2007

SAP and High Availability

High Availability (HA) is the concept of allowing several machines to work together as team. This team of computers, known as a cluster, helps each other out; they divide work amongst themselves, and if one fails for any reason, the others take over its work seamlessly and automatically. HA technology is still under development for both UNIX and NT platforms.

Both operating systems currently support an HA solution at the operating system level. That is, from the view of the operating system, HA can work. If one computer fails, the other computer can sense this failure, and take over the network ID, the I/O operations, and other operating resources of the failed computer.

However, there is currently no solution, for any platform on any hardware, that can offer an HA solution for applications such as SAP R/3. Although the HA solution at the operating system may be able to automatically make the resources of the failed computer available, it still cannot properly shutdown, roll back and/or roll forward, and restart the database and SAP processes. Thus, at present, if a productive SAP R/3 server fails, it always requires manual attention.

Although SAP, Microsoft, Oracle, and almost every major hardware company are currently working on a more complete HA solution, it will probably be a couple of years before a true HA clustering solution is available. A true HA cluster solution would allow a group of clustered servers to be viewed by the end user as a single server, and even when some servers fail, service is uninterrupted (though performance decreases). In this case, SAP on an HA cluster machines would look like this:

No comments: