Poor performance when archiving Rl_TA and RL_TB objects

[Former Member]

Hello Everybody:

We are in an Archiving Project and we had experienced a very poor performance when archiving RL_TA and RL_tb objects.

Thus we have to archive a lot of data, we had calculated that with the actual time of processing we will need almost a year to do this archiving data.

Does anybody have a tip or recomendation in order to improve the performance whe n archiving those objets?

We had look for SAP Notes about it without find something usefull

Thanks in advance for you valuable support on this matter.

Best Regards.

Julio Cesar

[JL23]

are you refering to the write step or to the delete step or to the store step?

what is your SAP release? Which report is used for archiving?

can you give an example for the poor performance? How many records do you have in LTAK, how much did you archive and what time was needed?

do you just archive one warehouse or many warehouses in one job?

do you have any exits with own checks in use?

how is your archving object customized? how much MB can the size be or how many records may an archive take?

when did you run the archiving? during the day while you had many movements in a warehouse, or as a nightjob when there is not much activity in the warehouse?

Having this info I could at least compare with my values and can tell you if it is a bad performance or just normal.

[Former Member]

Hi Jürgen:

Poor performance when archiving Rl_TA and RL_TB objects

Message: are you refering to the write step or to the delete step or to the store step?

Write 22 min

Delete 11 min

what is your SAP release?

SAP ECC 6.0

Which report is used for archiving?

Transaction SARA.

Objects RL_TA ( Standard Program RLREOT00SN)

RL_TB ( Standard Program RLREOB00SN)

How many records do you have in LTAK,

140,000 records

how much did you archive

Write 30.00 min

Delete 13.00 min

what time was needed?

lower to 10.00 min

do you just archive one warehouse or many warehouses in one job?

8-10 warehouses

do you have any exits with own checks in use?

NO

how is your archving object customized? how much MB can the size be or how many records may an archive take?

100 MB in the file archiving

total of records 3,510,000

when did you run the archiving?

In normal operation and at night

[JL23]

you did not say how many records you archived (wrote into the archive) in those 30 minutes.

[JL23]

This are my numbers:

I have 1283105 records in LTAK

my archive run is for documents older than 180 days and runs monthly

I have 41 warehouses in the selection

and archived (write job) 218259 records in 48 minutes in my last run.

the delete job took 10 minutes

I am using the old program RLREOT00S

Edited by: Jürgen L. on Feb 6, 2012 8:33 AM

[Former Member]

HI Jüren:

Sorry for the delay in answer you, but yesterday was an unlaborable day at Mexico

In RL_TA we archived 85,975 records in Write / Delete processes

In RL_TB we archived 28,232 records in Write / Delete processes

Thanks in advance for your valuble support and help

Best Regards

Julio Cesar

[Former Member]

Hi Jürgen:

We are using the new Program RLREOT00SN

Have you ever used the new Program?

Did you find that was better work with thw old Program?

Thanks in advance for you support an help in this matter.

Best Regards

Julio Cesar

[JL23]

the new archiving program came with OSS Note 1137327 - Writing archives for RL_TA and RL_TB:Performance improvement

and if I understand the note correctly, then it was because the old programs had performance issues.

If I compare my runtime with the old programs with your new programs, then I can see that I archive more records per minute than you do.

of course performance is dependend on many other factors. Not just on TO headers. If your TOs have many more items than mine, then it is not really comparable.

But I see actually no need for me to change to the new programs. My programs are scheduled and run twice a month automatically, already for many years without any issue.

And I dont see any bad performance in the given numbers, I think they are just fine compared to other archiving objects. I have jobs that run for many days. e.g. material master archiving needs about 4 minutes per material in average.(this is what I call poor performance)

Looking at your given record numbers you could archive all your TOs in 2 hours, I wonder what you calculated when you said:

we had calculated that with the actual time of processing we will need almost a year to do this archiving data.

[Former Member]

Hi Jürgen:

We think that we made a mistake when gave you the answer to your first question, and thas is the reason that you think that we can archive aour data in two hours.....

Then let me give you some information:

1) this will be the firts time that information will be archived

2) Now a day for RL_TA we have

Table LTAK with 22,620,329 records

Table LTAP with 31,080,990 records

3) For TL_TB Object we have

Table LTBK with 12,191,330 records

Table LTBP with 30,656,546 records

With these numbers I calculated the total time that we need to archiving those Objects

I hope this information let you undertand why we spoke about poor performance

Thanks in advance for you comments

Best Regards

Julio Cesar

[JL23]

okay you missed to archive your records for a long time, so you should start sooner than later.

But you still have a mistake in your calculation.

if you archive 89975 in 30 minutes, then it takes 125,7 hours to archive 22,6 million TOs, which is not more than 5 days.

This means it archives 50 TO's per second, not poor from my point of view.

And it is a one-time effort. once you are in a regular archving interval like i am, then you may only need a few hours every month. And if you archive daily then you will only have a few minutes, this should not harm your overall performance and background processes.

Furher you can setup multiple variants (e.g. per warehouse one) and jobs, so they can run in parallel.

[former_member223112]

Hi Jurgen,

Thanks for your information. This discussion was really helpful.

I have a query. I have 1.5 million records in  LTBP and 1 million records in LTBK.

Approximately how much memory can be saved?

Regards,

Srinivas

[JL23]

Here is a statistic for RL_TA from SARA:

And RL_TB:

Please do not reopen old discussions, see