How many testing resources & days are required for a project span of 6 months?
It is too vague a question to give a proper answer. You need to gather more information about the project to make a proper assessment.
You need to know the project size first.
1. Do the breakdown of the project into modules
2. Break the modules into functional points or scenarios
3. Apply the estimation for functional points based on complexity and dependencies
4. Apply contingencies and risks
5. Consider your company process oriented factors, etc.
Here are some of links I got easily that you can refer to get some idea.
How to do Estimating Testing effort ?
Software Test Estimation: An Expert View
Thank/Like to help others if my input helped you !!!
I am speaking for myself, not for my employer nor any one.
It depends on test plan and test strategy.
It depends upon many factors.One cannot specify a certain period without knowing the details.
We can't say until and unless we are not aware of testing process of the project and its details.Provide more details to get answer for your query.
As the other's have said, it's too hard to estimate without knowing all the details and doing your due diligence. I'll just chime in and add some tips on the sort of off the off the cuff thing you can use if the project manager puts a gun to your head and needs a quick answer.
Originally Posted by romeo_nayak
Typically, Engineering teams are staffed at a 3-1 developer to QA ratio for QA to keep pace with new development in a manual heavy shop, and 6-1 when there is a high level of automation project. You can probably expect around the same ratio in the amount of *new* testing work produced.
So facing a lack of information, I would use that 3-1 rule. For every 3 man/days of Dev work, I would expect at least 1 QA man/day in manual testing in either testing new features or writing automated test cases.
The 2nd part is, How much affected functionality is left to be tested as part of *regression*. Hopefully you'll have a TCM solution that has gathered run time statistics of your last 3 regression run so you can simply add that time to your estimate. If you have full regression automation, even better.
But if you're stuck with heavy regression without automation or base metrics, you could propose a time box regression method where you spend say 1 man/hour on each Feature/sub feature, where each of those features will be cursory tested within that timebox. That will give you a good amount of regression coverage while giving business an accurate time estimate they can go by.
If you survive all that multiple times, I would recommend learning Krav Maga disarming techniques.