-1

  1. Some people says buffer for the table will create issue while running a module in appserver may I know the reason.

  2. In my Main module I’m using For each ... break by condition to read the record from DB which needs to read 100k of records within minute. So if I'm not using Break By condition in the For each loop and running batch parallel (using asynchronous method based on input count will create multiple session) through appserver then I can able to improve performance little bit but i need to improve the performance even more what else i can do?Please suggest me..

eg: RUN report.p ON SERVER sh ASYNCHRONOUS SET ah EVENT-PROCEDURE "reportDone" IN THIS-PROCEDURE (input aaaa, OUTPUT yyy AS INT).

  1. Buffer vs Temp table for a table which will give high performance?
James Z
  • 12,209
  • 10
  • 24
  • 44
user3239566
  • 5
  • 3
  • Progress 4GL does not support multithreading so your "running parallely" comment isn't making a whole lot of sense. You might want to consider posting some sample code to clarify your question. – Tom Bascom Apr 18 '14 at 18:39
  • using asynchronous method based on input count will create multiple session eg: RUN report.p ON SERVER sh ASYNCHRONOUS SET ah EVENT-PROCEDURE "reportDone" IN THIS-PROCEDURE (input aaaa, OUTPUT yyy AS INT). – user3239566 Apr 19 '14 at 19:04
  • Are you running more than 1 async app server process? It is very unclear what you are doing and what you are asking. What any of it has to do with buffers and BREAK BY is a mystery. – Tom Bascom Apr 19 '14 at 20:16
  • When increasing performance you should always check that you have high buffer hits and smart logic. Post code and schema of relevant tables! – Jensd Apr 22 '14 at 05:39

1 Answers1

0

It makes no performance difference if you reference a defined buffer or directly use the table name (that just results in the use of a "default buffer").

The use of temp-tables depends entirely on how you go about it. There will be overhead in "marshalling" records into the temp-table originally but there may be sufficient processing advantage to offset that. But there is no one size fits all "which is faster" answer.

Tom Bascom
  • 13,405
  • 2
  • 27
  • 33