The heat and stress testing of computer components in the lab does not necessarily bear out how components will behave in the field, according to a study done by Google. When you are Google, and you have millions of server nodes in production using a mix of different technology, you can actually study component failures with a …
Errors, errors, everywhere...
Having once developed and supplied hardware error correcting memory systems to a number of nuclear power stations, and having also monitored the errors rates, there were significant differences between sites though there was no obvious correlation with hardware batches. We put it down to variations in background radiation - maybe Google should look at their local geology and radon etc. levels.
Th@t's od^, I'vE* never had a mEmoy corrup&^)*(&^&(*
Be aware of how Google purchases memory
As I understand it Google purchases the majority of it's DIMMS from stock which has already failed QA at manufacture time, the logic being that even if 75% are no use it can still be cheaper if you buy millions of DIMMS in bulk.
I would be astonished if this study didn't show higher than expected error rates, previous studies would I assume be on DIMMS which weren't dodgy to begin with.
"As I understand it Google purchases the majority of it's DIMMS from stock which has already failed QA at manufacture time..."
If you don't have a source, I'll presume you're incorrect -- at least until you learn how to use an apostrophe correctly.
Cosmic Ray hits
I think Julian has one of the keys.
I recall early chip makers were plagued with unexplained errors which were found to be radiation induced.
Thanks, Google, for sharing some very interesting info.
Paris because she causes memory errors.
Google has mostly small servers wit a modest amount of memory/number of DIMMs in each small server?
nice to know I share old out date technology with google
nothing wrong with out of date tech if you've got enough of it!