MVSFORUMS.com

[Mukunda]

Gurus of the board
We did some performance tuning for a lengthy batch job and reduced CPU cycle time by 90%. Now the management is discussing on whether to expand this to the whole application? Is there a dollar incentive on reducing CPU cycles for batch applications?
Our shop owns the hardware (not renting it).
Thank you!

[kolusu]

Mukunda,

Since you own the Hardware, there will be no Monetary gains for you. You can expand it to entire application provided that you have manhours for analysis and fixing. This will helps to have Smaller batch window. ie. total time to complete the entire batch cycle is reduced. May be you can sell saved time frame for lease and there by making money on it.

On a side note you can use this program as an example and create a standards document for developing highly efficient programs.

Kolusu
_________________
Kolusu
www.linkedin.com/in/kolusu

[Mukunda]

Thanks Kolusu.
Reduction in the batch window and associated reduction in monitoring time were the benefits we were envisioning.
Besides, there was a thought that if the CPU gets free'ed up it might delay the addition of more processing capability (hardware). I guess it depends on several other factors. But is that something worth looking at?

[superk]

The first shop that I worked in was a bank (in the US). At that time, the state we were based in taxed all corporations on the number of CPU cycles they used per month. Later, when we merged with another bank, that was the main reason Management gave for their decision to abandon that Data Center and to move it to another state.

[semigeezer]

Wow! Talk about government out of control!

The only major easons to save CPU cycles are time and money. If neither matter, or if you have a box with so much excess capacity and no plans for any future work to be added, then don't worry about it. But if you live in the real world, that box will have more work, the cost of software will eventually go up, the price of running the system will rise or some other major environmental changes will occur and your wasteful programs will start to cause problems. I've seen hardware vendors sell new boxes already anticipated to go into production at 95% capacity!!! Every lost CPU cycle on the new box could cause system wide impact.

But to the original question, the only reason to rewrite the code is if the cost of keeping it the way it is will be higher than rewriting it in the long term. (again, cost is measured in money, time, work spent fitting other programs to run with the current stuff, hardware requirements, etc).

[Mukunda]