There was a silly article in a recent copy of EDN talking about techniques to extend the lifetime of EEPROM for high event frequency datalogging applications.

It wins my BOZO THE CLOWN AWARD.

The article goes to great lengths to explain a paragraph worth of a technique sometimes referred to as rolling records. His techniques were simplistic and lacked the refinement that seasoned use would achieve.
However, it does introduce a concept that has been asked several times on this forum: Can you use EEPROM as RAM? Read the article if you are unclear about this topic.

In my lab, the author would have been fired for not realizing the PROPER business solution. Here is the basic problem he posed:

A device must datalog and timestamp a record once a second when powered. Records must be placed in EEPROM and be able to function 5 years at 8 hours a day. The timestamped info comes from an self-battery-powered RTC. <DUH ALERT> The article discusses elementary techniques for getting beyond the write cycle lifetime of EEPROMs. (good conceptual introduction)

He's why the author was fired hypothetically:

Most RTCs have a little battery backed RAM for his non-volatile data pointer. If his RTC doesn't, use another one.

There are much better techniques for rolling records if you didn't have the RTC battery backed RAM available. I could tell from the author's progression that he was going to suggest a dumb process before he wrote it. The erasure of a page is what I expected him to suggest and he did. Another silly viewpoint from someone that doesn't understand process efficiencies.

His techinique besides being lame and wasteful, is suspectible to being clitched if powered down during that write. Only a single byte write is necessary for what he intended. Had he worked more with circular buffers, the author might have seen the right solutions.

aka J