TREX consists of the following components
JAVA client
TREX provides several interfaces that can be used to integrate TREX functions into an application. The Java client is an interface that Java applications can use to access TREX. The Java client is integrated into Content Management. This means that the TREX functions are available in Content Management and therefore in the portal.
Web Server with TREX Extension
Content Management (more precisely, the Java client) accesses the TREX functions using aWeb server. Communication between Content Management and the Web server takes place using HTTP/HTTPS and XML. The Web server receives requests and forwards them to the index server and queue server. The servers then process the requests. A TREX component that enhances the Web server with TREX-specific
functions is installed on the web server. On windows this component is realized as as an ISAPI server extension for the Microsoft Internet Information Server.
Queue Server
The queue server enables the asynchronous indexing of documents. It has a separate queue for each index. It gathers documents to be indexed into one of the queues. It transfers documents to the index server for the actual indexing process at regular intervals. You can use the queue parameters to control when and how many documents are transmitted. This allows you to schedule indexing for times at which the index server does not receive a large amount of search requests. The queue server forwards the documents to the preprocessor before transmitting them to the index server.
Preprocessor
The preprocessor has two tasks:
1. When TREX processes search requests, the preprocessor carries out a linguistic analysis of search queries. The preprocessor passes the results of the analysis to the index server, which then processes the query further.
2. When TREX receives documents to be indexed, the preprocessor prepares the documents for the actual indexing process.
Name Server
The name server is used with large distributed TREX installations. It uses its databasis to store and coordinate system-wide information. It also ensures that the TREX servers can communicate with each other and that TREX can communicate with Content Management. The name server is also responsible for distributing the system load if more than one TREX server is capable of carrying out a task. In a distributed scenario, you can install several name servers to ensure that a name server is always available. A replication procedure ensures that the databases of the different name servers are synchronized.
TREX service
After the TREX installation, a TREX service (daemon on UNIX platforms) runs on each host. The TREX service is a central service that starts the actual TREX servers (index server, queue server, and so on) and monitors them during routine operation. If a server
becomes unavailable, it is automatically restarted by the daemon. In the standard configuration, the TREX service starts all servers that have to run for a minimal TREX system. For a distributed installation, you can use the service configuration file to control which servers the service starts on individual hosts.
Python Support and Python Extensions
Python support includes scripts that are used to test TREX functions. For example, you can test creating indexes as well as indexing and searching documents. Python support also includes a tool for administrating the name server (name server administration). Some name server functions can only be administrated using this script. Python extensions are scripts that enhance standard TREX functions with optional features. Index replication is one of these optional features.
SAP Trex information
