It probably depends a lot on if you are ultimately going to be trading an automated (or at least mechanical) strategy. In that case nothing is stopping you from backtesting several years worth of data (provided your trading platform allows for that).
On the other hand if you will be trading a discretionary strategy, well then I guess it becomes more discretionary Personally, I have been forward testing my discrectionary strategy for the last 4 months and will continue to do so till the end of the year. In January I plan to take it live at which point I would have forward tested it for over 6 months.
Before my current strategy which essentially revolves around PASR, I was trying to create a mechanical system. I forward tested that for 6 months before ultimately deciding to drop it. I do not regret those 6 months or think of them as wasted because without them, I wouldn't have gained the knowledge to progress to my current strategy which has performed well enough over the last few months for me to be willing to take it live in January.
I guess 6 months worth of forward testing may not be necessary for all traders, but for me it has allowed me to get a feel for it through various trading environments and importantly, allowed me to build up statistics for it. If I had only tested it for 2-3 months, I might trust it less because I dont have the stats to back up it's profitability.