[Gllug] NFS problem

[James Roberts]

Anthony Newman wrote:
> Kostas Georgiou wrote:
>> All[1] SATA disks have an uncorrectable bit error for every 10^14 bits
>> read, with 1TB disks close to 9*10^12 bits you have ~9% chance to get an
>> error when you read all the data. 
> 
> One would presume that the BER is largely independent of disk size and 
> if anything will have improved with time, so the total array size is the 
> only important factor. I haven't heard any doom-mongering about the 
> potentially gigantic SANs that large companies have been using...

Multi-terabyte SATA/SAS SANS (which we sell) tend to come with RAID6 
capability to obviate this problem when configured with SATA. The error 
rate (and array size) is much smaller with SAS.

There's lots of interesting stuff on the Adaptec forums re this, e.g:

http://storageadvisors.adaptec.com/2005/11/01/raid-reliability-calculations

There had used to be a link to a spreadsheet for calculating the failure 
rate and error certainties in big boxes but it is now missing...

MeJ
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug