Loading In-Memory Tables
Tables are loaded into TIBCO Patterns products in two ways:
| • | From a remote source (such as a DBMS) via a TCP socket connection to the TIBCO Patterns server, or |
| • | From files local to the host that is running the TIBCO Patterns server (server-side loading). |
This section describes only the common features of loading tables, regardless of the method used to manage this process.
The TIBCO Patterns server optimizes the loading of a table based on the number of records it expects. When more than 40,000 records are to be loaded into a newly created table or added to an existing table – the TIBCO Patterns server employs multiple processors to speed up the loading. (For smaller loads, a low overhead, the single-stream load is performed.)
Loading From Files - Local File System
TIBCO Patterns also comes with easy-to-use utilities for reading client-side data files and preparing the records for transmission to a TIBCO Patterns server. These utilities support comma-separated values .csv files.
The caller can specify an estimate of the number of records to be sent. Based on the estimate, the TIBCO Patterns server uses multiple processors or a lower overhead single processor load. If an estimate is not given, it assumes that a very small number of records are to be loaded and uses the single-processor low-overhead load.
Loading From Files - Server File System
The server can be instructed to read data from files on its file system. The location of these files is restricted and only files in the server's loadable-data directory can be read.
When loading data directly from a .csv file, the TIBCO Patterns server assumes a large number of records are to be loaded and employs multiple processors to hasten the load.