ABAP new Open SQL and CDS runtime
With the introduction of NW AS ABAP 740 SPS5, the New enhanced extended Open SQL was introduced. The new enhanced open sql has some good features such introduction of aggregate functions, right outer join, casting, etc making it a starting step for Code-Push down paradigm.
The Code push down techniques (CDS and New SQL) are compared in the demo here. Below is a demo program of data retrieval from BKPF and BSEG.
1) ABAP program using inner join between BKPF and BSEG
2) ABAP program using CDS view entity
3) ABAP program using enhanced Open SQL features
So its Old SQL vs CDS views vs New Open SQL
The objective of this demo is to determine the run time difference in microseconds.
1) Lets start with the ABAP program using open sql syntax and check the run time.
The Output
2) Next, lets move to program using CDS views. Not going to elaborate on the CDS views as the intention here is to measure the run time.
CDS Views
The ABAP program using the CDS view
The Output
3) Lastly lets move to program using enhanced open sql syntax.
The Output
Here is Run time comparisons in micorseconds.
Since the system resources load differ at times, multiple run times are recorded.
These findings may vary depending on the system load. Between CDS and New SQL, both stand atop for the new code paradigm.
From the Code-push-down perspective, CDS and New SQL refer to the Top to Bottom approach. However there is Bottom to Top approach which is via HANA views.
The screenshot suggests that ZDEMO_CODE_PD_OPEN_SQL only uses one join condition for the inner join, while the other examples use three conditions.If your screen shot is accurate I suggest to add the missing conditions and repeat the measurements.
Thanks for the useful document.
Also selects 10 records instead of 100 from the other examples
CDS is considered good for reuse scenario. And not to forget buffering annotation available in CDS which has potential to dramatically improve reuse cases.
Thanks,
Sumit