There are various ways of structuring distributed systems. The table below contains an overview of the systems that are supported. Because the index servers are the central components, the systems are classed by the role of the index server.
Supported Systems
Number and Roles of Index Servers | Data Storage | |||
---|---|---|---|---|
Backup | Master | Slave | Decentralized | Centralized |
- |
1 |
1 |
! |
! |
- |
2 or more |
At least 1 per master |
! |
! |
1 |
1 |
1 |
! |
|
1 for all masters |
2 or more |
At least 1 per master |
! |
|
1 per master |
2 or more |
At least 1 per master |
! |
SAP recommends configuring at least two slave index servers for each master index server, to noticeably improve the performance of the TREX search. However, all other combinations of master and slave index servers are also possible (for example, one master and three slaves). You can also start with a minimal configuration of one master index server and one slave index server.
The following is valid for all supported systems:
The graphics below depict systems on individual hosts. However, all graphics are also valid for blade systems.
The graphics below depict systems in which Web servers and RFC servers run. If only one type of connection is relevant, only Web servers or only RFC servers can run.
The sections below describe the supported systems in detail. In the details, the recommended ratio of one master index server to two slave index servers for improved TREX performance is assumed.