Archives of the TeradataForum

Message Posted: Tue, 07 Apr 2004 @ 02:46:38 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

On Windows (XP/W2K/NT), it is a two phased approach. The first phase is to start the load utility (MLOAD/FLOAD/TPUMP) with the named pipe access module np_AXSMOD. The access module will create the pipe file and will pause at the beginning of the data acquisition phase awaiting input from an external data source to the pipe. The file must be specified by the root path "pipe" in the file name in order for the access module to create a virtual pipe file.

The second phase is to start the FEXP (or other process) and simply write to the same pipe file name specified after the pipe reading utility has reach acquisition phase and awaits data (you can see the empty initial fallback file created below in the fallback file path).

The parameters to the access module are as follows:

np_axsmod parameters (all space separated)

ld - named pipe log directory path

fd - fallback file path

ff - fallback file name

cfd - confirm fallback file delete (y/n)

b - blocksize (max 64K)

ll - log level (1 - 6) Warning: Log Level 6 logs each IO.

Example MLOAD

The process can specify different servers and the following Windows UNC assumes the current server name (<\\.\> <<file:///\\.\>> ) for both processes, but again you must specify "pipe" in the root path to create the virtual file.

     .IMPORT INFILE <\\.\pipe\mypipename> <> ;

         AXSMOD np_axsmod 'ld=d:\nplogs fd=d:\npfallback =
     ff=mypipename.fb
     cfd=n b=65477'

         LAYOUT Layout1

         APPLY INSERTS ;

Example (FEXP) - executed after MLOAD reaches acquisition phase

     .EXPORT OUTFILE <\\.\pipe\mypipename> <>

          mode indicator

          format fastload;

Kyle Prescott

Teradata Certified Professional

UnumProvident - Data Strategies, 3-South

	<Prev	Next>		<<First	<Prev	Next>	Last>>