2504657 - CSV Input Adapter - Strange behavior with 160 columns - SAP Event Stream Processor

• The SAP Event Stream Processor (ESP) project fetches data from a CSV file.
• When the CSV file has 160 columns, it is observed that the throughput performance degrades to only 35 – 40 records/second.
• However, when the CSV file gets modified to have one less column, ie 159 columns, much better throughput is observed, of 18,000 - 20,000 records/second.
• When testing to see if the problem is caused specifically by the data in column 160, it is observed that the issue is only when fetching all 160 columns.

For example:

• Scenario 1: When removing the last column (Column_159) and deploying the project, high throughput is observed (~18k - 20k).
• Scenario 2: When removing the first column (Column_0) and deploying the project, high throughput is observed (~18k - 20k).
• Scenario 3: When removing a random column from the available list (Column_88), again, high throughput is observed (~18k - 20k).

SAP Event Stream Processor (ESP) 5.1

streaming, processing, esp , KBA , BC-SYB-ESP , Sybase ESP (Event Stream Processor) , Problem