Originally posted by: paul.brett
When you deploy a map into Sterling Integrator it's important to know that any settings that create files that use 'Map' location will result in files being created in the SI/install/logs folder.
In other words, when the file location setting for the Workspace is set to 'Map', the folder used is the SI/install/logs folder.
So you need to make sure that when you run large transactions, the space allowed in the SI/logs folder filesystem is enough for the Workfiles to be created.
Also when you are running large transactions with multiple records, I would try to execute the map by bursting the inputs so we work on the transaction chunk by chunk and thus avoid this critical memory usage. Edit your input card settings.. 'Fetch As', and set it to 'Burst' instead of 'Integral'. Then set a number of records to fetch in each burst. If for example, you have 2,000,000 records to read, I would set the Fetch Unit to 100,000 so the map will burst 20 times.
See the documentation about burst here:
http://www.ibm.com/support/knowledgecenter/SSVSD8_8.4.1/com.ibm.websphere.dtx.md.doc/concepts/c_map_design_Using_Bursts_for_Large_Files.htm
http://www.ibm.com/support/knowledgecenter/SSVSD8_8.4.1/com.ibm.websphere.dtx.md.doc/concepts/c_map_design_Configuring_Bursts.htm
You should also tune your Workspace settings for the map, using the DTXPAGE tool, to find the ideal values, based on your input.
(with thanks to Ejay for the hints)
Thank-you.
Paul.
Follow me on Twitter
#IBM-Websphere-Transformation-Extender#IBMSterlingTransformationExtender#DataExchange