Philippe-FR
Member
Hello,
first of all I apologize for the probably stupid question I will ask here. I have to deal with a JDE E1 system in the company I work for. We have performances issues and one possible solution is called "subsystem" but we can unfortunately get a clear answer on the limits of this.
Here is the situation : we have 23 warehouses where people are selling goods to cutomers. Obviously, they have to print invoices. For this, 23 versions of the R42565 program have been created. All these version are run on the same and mono-thread queue. As time goes by during the day, the "waiting time" in queue of each submitted job increase. At the end of the day, the time it takes to get the invoice can be more than half an hour.
Question 1 : R42565 seems to be a program that, according to the litterature we have found, the guys we asked to and the Oracle support, must be run on a mono-thread queue. It's true that we have had some bad experience in the past, running multiple R42565 at the same time ending in mixed informations printed on the invoices. It is also said that "you can run multiple R42565 if they use different data selection". The question I never got a clear answer to is : "can we run multiple R42565 jobs (with different data selection) on multiple single threaded queues or do we have this job only once at a time on the entire system ?". In the first solution, the idea might be to have for example 23 mono-threaded queues. We could then run up to 23 R42565 jobs at the same time, but at any time, only one job of a specific version would be running. You see what I mean ?
Later on, someone told us "why don't you use subsytems ?". Great idea, but once again, we can not make a clear idea of how to set up all that properly.
If I try to compare this to some real life situation, I have found that subsystems compares to business class desk at the airport when you register for a flight : if you have a business class ticket, you don't have to queue with people travelling in economy, you have a special desk. The idea is that both desk are doing the same thing (they register your cases) but you have some special element that allow you to use one queue instead of the other.
Question 2 is : we have 23 warehouses. We have created 23 R42565 versions, each version specifying the warehouse id it is intended to. Could we create 23 subsystems to run on each a specific version so that each warehouse would be able to print invoices without any delay ? what about the multiple R42565 jobs running at the same time (cf question 1 above) ?
Once again, I apologize for the question, wich is certainly due to a lack of knowledge on JDE. Unfortunately, among all the (said) experienced (and exepensive) experts we talked to about this, none could give us a clear and efficient answer ...
Regards,
PhC.
first of all I apologize for the probably stupid question I will ask here. I have to deal with a JDE E1 system in the company I work for. We have performances issues and one possible solution is called "subsystem" but we can unfortunately get a clear answer on the limits of this.
Here is the situation : we have 23 warehouses where people are selling goods to cutomers. Obviously, they have to print invoices. For this, 23 versions of the R42565 program have been created. All these version are run on the same and mono-thread queue. As time goes by during the day, the "waiting time" in queue of each submitted job increase. At the end of the day, the time it takes to get the invoice can be more than half an hour.
Question 1 : R42565 seems to be a program that, according to the litterature we have found, the guys we asked to and the Oracle support, must be run on a mono-thread queue. It's true that we have had some bad experience in the past, running multiple R42565 at the same time ending in mixed informations printed on the invoices. It is also said that "you can run multiple R42565 if they use different data selection". The question I never got a clear answer to is : "can we run multiple R42565 jobs (with different data selection) on multiple single threaded queues or do we have this job only once at a time on the entire system ?". In the first solution, the idea might be to have for example 23 mono-threaded queues. We could then run up to 23 R42565 jobs at the same time, but at any time, only one job of a specific version would be running. You see what I mean ?
Later on, someone told us "why don't you use subsytems ?". Great idea, but once again, we can not make a clear idea of how to set up all that properly.
If I try to compare this to some real life situation, I have found that subsystems compares to business class desk at the airport when you register for a flight : if you have a business class ticket, you don't have to queue with people travelling in economy, you have a special desk. The idea is that both desk are doing the same thing (they register your cases) but you have some special element that allow you to use one queue instead of the other.
Question 2 is : we have 23 warehouses. We have created 23 R42565 versions, each version specifying the warehouse id it is intended to. Could we create 23 subsystems to run on each a specific version so that each warehouse would be able to print invoices without any delay ? what about the multiple R42565 jobs running at the same time (cf question 1 above) ?
Once again, I apologize for the question, wich is certainly due to a lack of knowledge on JDE. Unfortunately, among all the (said) experienced (and exepensive) experts we talked to about this, none could give us a clear and efficient answer ...
Regards,
PhC.
