Seagate ST31500341AS 1.5TB

Just chatted with a Seagate rep, and he said that a few initial Reallocated sectors is "normal" (under 100 total). Here is the chat transcript:

I have 3 Seagate ST31500341AS drives I recently purchased (all CC1H firmware) which are showing SMART Reallocated Sectors. ... Jeff W: We can always repalce the drives as long as they are in warranty Me: so the drives should never be reallocating sectors? Jeff W: But I would like to make sure it isn't the raid controller causing the sectors Me: looks like 1 drive has 6, the other has 9, and the 3rd has 0. Jeff W: Under 100 you are fine Me: so a few realloc sectors is normal? Jeff W: yes Jeff W: That is why we add them to the drive Jeff W: as long as you are not near 100 you should be ok ...

Not sure if this is correct or not as the number of realloc sectors he stated (100) seems a bit arbitrary, but it's what Seagate reps are saying.

Anyhow, I came across a Google's study on SMART Reallocated Sectors performed on over 100,000 drives which is sort of interesting as well (see "Conclusion" section of the PDF at http://labs.google.com/papers/disk_failures.pdf):

[Google's] results confirm the findings of previous smaller population studies that suggest that some of the SMART parameters are well-correlated with higher failure probabilities. We find, for example, that after their first scan error, drives are 39 times more likely to fail within 60 days than drives with no such errors. First errors in reallocations, offline reallocations, and probational counts are also strongly correlated to higher failure probabilities. Despite those strong correlations, we find that failure prediction models based on SMART parameters alone are likely to be severely limited in their prediction accuracy, given that a large fraction of our failed drives have shown no SMART error signals whatsoever. This result suggests that SMART models are more useful in predicting trends for large aggregate populations than for individual components. It also suggests that powerful predictive models need to make use of signals beyond those provided by SMART.

I'm not an HD expert, but I'm wondering how much to trust low-volume SMART errors? Are they really conclusive as a measure of drive health (since I don't want to RMA a thousand drives to find one without any SMART errors if it doesn't really matter)? ugh...