Extract:
In a real-world production environment, the distributed systems of SAP Hana are common, because as data volumes expand, often single-machine sap HANA tends to become more and more strained, so it's necessary to build a distributed system that distracts query pressure and keeps SAP Hana high-speed. On the other hand, in the actual production environment, in order to prevent single-machine failures, HA is usually done so that the system can remain stable after a node host is down.
The above-mentioned SAP HANA Distributed System and high availability of the actual needs, but limited by the test conditions, the need for more than one SAP HANA Server, the General people do not build their own distributed Hana, most of them are on a single test. But the distributed system of SAP Hana is a little bit different from the distributed system we normally remember, and the distribution in SAP Hana mainly refers to ' memory ' distributed, not data file distributed.
For multi-master SAP HANA Systems, mentioned earlier, is a solution based on shared storage. But what are the specific solutions? Currently included solutions include NFS, GPFS, Storage Connector API. If you need to build a distributed SAP HANA system, building a shared storage system is the first step to make.
NFS: Easy to build, but poor performance, the test environment can
GPFS:IBM's high-performance file system, good performance, but complex to build, generally recommend the use of IBM Remote Engineer Services
See here: SAP Hana Distributed Systems and high availability (i)