CERN publication on bit error rate occurance

[stalni]

As the topic title says, I have some how, don't remember where(maybe HSS forums, could not find it when searching) found this document, outlining the modern(in 2007) problems in that occur when dealing with large amounts of data.

They write that modern data centers(there case is CERN located in Switzerland) need to take into account the failure rate(bit errors are there focus) of there hardware, and conclude that bit errors are to be expected, and therefore they need to implement solutions to detect and resolve the bit errors.

 

The document is provide mostly for fun, but of course when you are talking about RAID to solve all of your HDD failure needs(do realize that most of you make regular backups), I feel after reading this document that RAID is not sufficient for me:)

 

Now for anyone reading this document(it is 7 pages), before getting all worried about your data, please do take into account the amount of data you typically would deal with.

 

Please do fill out your opinion of the document in the poll

[tojoski]

The bit-error rate comes into play more than most people realize, especially when you start pushing 10tb+ arrays..

 

Its astounding to me that Microsoft is so far behind in this area.. No checksummin, copy-on-write or snapshots.

 

This is why I would eventually move to a zfs backend for my storage.

[dvn]

Sounds like one of you guys needs to jump onto one of the podcasts.

[stalni]

The bit-error rate comes into play more than most people realize, especially when you start pushing 10tb+ arrays..

 

Its astounding to me that Microsoft is so far behind in this area.. No checksummin, copy-on-write or snapshots.

 

This is why I would eventually move to a zfs backend for my storage.

 

I agree, with the long term move to zfs or similar systems.

 

On another note, I guess my poll didn't work out:( as I can not see the option in my post.

or is it just me?

[stalni]

Reupload of original file:

Data_integrity_v3