Raghunathan:
You may want to know I have implemented quite a few parameter mechanisms that used serial eeprom for the storage of the values. I have many times totally simplified the design of the user interface (and maybe saved a button in the process) by getting rid of the enter or acknowledge button. Instead at input of UP button I would read the current parameter from eeprom step it up one notch and then write it back. Similar for the DOWN button read the current parameter step it down and then immediately write it back to the eeprom. In general I would challenge anyone to sit and press the up and down buttons on a simple product two or three hundred thousand times. It is also not uncommon to find that tact switches themselves are only rated for reasonable service life of fifty thousand cycles (like this typical Omron part)

So by all means do evaluate the number of times you are likely to hammer specific eeprom or flash locations as part of your algorithms. But you are likely to find, just like I have, that wear leveling of non-volatile storage and the complexity that it adds is only required in certain cases where non-clever algorithms hit the eeprom or flash like an insistent jack hammer. Even for auto-update algorithms that are designed with a small amount of RAM based data accumulation and MCU controlled power-off it is possible limit eeprom updates to once in say ten minutes and not approach the lifespan of the most poorly rated eeproms in the operational life span of a typical product.

Michael Karas