Question I Need Readprobe.p Program

[Luciano Piérre Schulz]

Hi, somebody can send me readprobe.p program please?

thanks

luciano.pierre@totvs.com.br

 

[Cringer]

What is readprobe and what are you hoping to achieve with it?

 

[Luciano Piérre Schulz]

Progress 10.2B
OS Linux RH 6.8
I would like to get some information about performance of cpu and disks. Can the readprobe give me this results ?
I need some program to do a benchmark.

 

[Marco Mendoza]

... from old peg ...

 

[Luciano Piérre Schulz]

thanks Marco.

 

[cj_brandt]

readprobe was to measure reads from database buffers. the atm benchmark was to stress disk io.

 

[TheMadDBA]

Keep in mind that readprobe is a very specific test designed to test one thing. It is utterly meaningless as a benchmarking tool.

 

[TomBascom]

"Utterly" is kind of strong

If the idea is to get "one number to rule them all" then readprobe, atm or whatever are not appropriate. They are quite artificial and will behave very differently from your real code.

They are useful to help you understand how much room for improvement you might have in your code or to compare certain options that you might be looking at in your hardware.

In many respects they are both very simple tools. When used properly and the results interpreted with appropriate caveats they can be helpful.

 

[TheMadDBA]

I did say as a benchmarking tool

Obviously it is a fine tool for massive reads/scans. I just don't consider tests without writes, networks and the appropriate OE components a benchmark.

 

[TomBascom]

Since I am in one of those moods I will quibble and say that I certainly agree that neither readprobe nor atm nor any other tool that I am aware of are a comprehensive geenaeral-purpose benchmark for OpenEdge applications

The nature of benchmarking tools is to take something complicated and over-simplify it. Which is where the danger lies. If the user doesn't understand the particular compromises and simplifications being made and how they relate to the real target then it is easy to draw all the wrong conclusions.

 

[TheMadDBA]

I completely agree Tom.. that was the point I was trying to make in my first post. Most likely too brief/terse due to lack of sleep and coffee because of my company chosen (cheap) late night flight back home from the PUG

I didn't mean to throw stones at readprobe... just point out that if you want to benchmark (in the traditional sense) that you have a lot more work to do than that.

Edit: I am partial to FIO for general testing of disk IO. Same disclaimer applies.. you need to know your application and how OE reads blocks to get anywhere near to a valid test.

 

[TomBascom]

I think we are in "vehement agreement"