Performance difference BDC CALL TRANSACTION 'FB01' between ECC and S/4 hana

[former_member806206]

Hi, experts.

Now I'm change of BDC program conversion ECC to HANA, everything is good but except performance of CALL

TRANSACTION 'FB01' .

I found out S/4 takes more net times on CALL TRANSACTION 'FB01' than ECC..

Analyzing on SAT,

ECC - gross(mircosec) : 3,127,951 net(mircosec) : 7,752

S/4 - gross(mircosec) : 2,180,088 net(mircosec) : 233,046

Under same conditions(same code source, number of account documents header and item, dynpro screen etc).

Please tell me the reason and how can i optimize this?

regards.

[nomssi]

Hi Minjae Kim,

No explanation from me, have you repeated the measurements? Are you using customized measurement variants with different settings (Mesurement tab. -> Settings) ?

I would try comparing the measurements in SAT to get some hints:

• From the Evaluate tab you can download/upload a measurement (Export and Import) or use Send measurement by RFC to copy one trace result to another system.
• Once you have both measurements in one system (make sure to reset the default filter for user in the Existing ABAP Runtime Measurements list) you can select both measurement and use Compare Hit Lists.

I hope this helps to identify the root cause.

regards,

Jacques Nomssi Nzali

[former_member806206]

Hi Jacques.

thank you for your reply.

as like your said, I already done comparing hit lists between ECC and hana but there are no difference with call transaction hit lists..

now i just doubt wether CALL TRANSACTION 'FB01' of S/4 hana have bad performance than ECC..

[Sandra_Rossi]

3.1 seconds in S/4, 2.3 seconds in ECC. Different system, different performance. Nothing abnormal.

NB: about "I found out S/4 takes more net times", I think you don't understand what is the difference between gross time versus net time, because in the context of your question it's meaningless. I guess the right measurement is 3.1 seconds versus 2.3 seconds.

[former_member806206]

Hi, Sandra Rossi.

thanks for your reply.

I would like to compare background job performance between ECC and S/4 hana so focused on net time.

plz tell me if my thought is wrong.

and one more question i wanna ask to you if you are knowledgeable about this, anyway the analysis with sat is like below

ECC - gross(mircosec) : 3,127,951 net(mircosec) : 7,752

S/4 - gross(mircosec) : 2,180,088 net(mircosec) : 233,046

but the duration of time with background job is like below,

ECC - 1 sec

S/4 - 3 sec.

I would grateful to you if you can explain what is reason background job take less time than S/4 in spite of taking more time with SAT analysis..(of course use same data record file with SAT analysis)

[Sandra_Rossi]

You're wrong.

GROSS time of a procedure = time that it takes

NET time of a procedure = time that it takes without considering the time of all "sub" procedures it calls.

So you should theoretically take the gross time. Anyway, you just want to measure the total time, so just look at the total time.

Please search the forum if you still don't understand.

[FredericGirod]

Even in the same system, same version you could have values like that if the stat, index are not up-to-date

[former_member806206]

Hi Sandra,

thanks for your reply.

I understand should with gross time of procedure even i wanna compare background job time.

thank you.

regard.

[matt]

Is it on the same hardware?

[former_member806206]

Hi, Mattew.

Of course did with same hardware..

[matt]

There's no "of course" about it. Upgrade/migration projects often take the opportunity to upgrade the hardware.

Anyway, taking 10x the time is a performance issue. Taking 1.5x the time (especially for a short response time, which has greater uncertainty), isn't.

[former_member806206]

Hi Matthew.

thanks for your reply.

As you said, taking 1.5x time with short time response is uncertain.

I would try other test case.

thank you.

regard.