Metrics on database read/write/delete based on size of the table

Question

Hi,
Though we have many performance measurement tools,sometimes it is difficult for developers to trace and measure each read, some reads which look fine in the development environment turn out to be show stoppers in quality environments.
I am trying to find if we can give a rough estimate of ideal response time of an RFC based on the number of database fetches/writes in the RFC assuming that the loops and internal table reads etc are optimized.
e.g.: if my RFC performs two reads and one insert.I would like to arrive at a figure e.g.: 200ms should be the ideal runtime of the RFC.
I would like to base my calculations based on the following parameters:
- Table Size
- Key/Index used
e.g.: For a FETCH operation
    Table Size | Key Used | Total Time 
     Upto 1 G  | Primary  | 100ms
     1 - 5  G  | Primary  | 200ms
Similarly for insert and delete..
I have the following questions for the forum in relation to the above:
- Is the above approach good enough for arriving at a
  approximate metric on total response time of an RFC ?
- Are there any other alternatives, apart from using the
  standard SAP tools ?
- How are metrics decided for implementations with Java
  and  .NET frontends ?
Thank you,
Chaitanya

Answer 1

Hi There
Do you mean dba_segments table?
My boss want to export 2 big tables and import to training environment, each table contains more than 2 million rows.
I want to know how big(bytes, or megabytes) are those two tables on the hard drive, because we are going to run out of the space on the same server. I am not sure the diskspace can afford such big export or not, so if I can know how big are those 2 tables, and then I can decide what I can do for export. For example: I got 200MB left on my /home directory, that is the only place we can put export, those 2 tables could be bigger than 400MB even though I compress the export file.
Hopefully this time it is clear.