Program Optimization

Beta rules of program optimization: run the program many time and watch closely

- time it - watch it's IO / CPU / memory usage. - watch it's file open()s. - have tea.

Also: Write down in English what it is you are trying to do. Is it doing what you think it's doing?

• lots of data lots of reports
• lots of data one report?
• lots of reports one dataset?
• batchable? threadable? increamental? can parts of the report be run on their own and then added together at the end?
• Do you have to do all the reporting at the same time ? Could you do some of the data every day such that at the end of the month all the * required reports are done? Rolling reports?
• How long should it take? and a year from now how long should it take?
• What is your budget for computing resources? and a year from now what sort of budget will you have?