[CentOS] Re: Anaconda doesn't support raid10

Feizhou feizhou at graffiti.net
Thu May 17 03:26:19 UTC 2007


Lanny Marcus wrote:
>> Message: 31
>> Date: Tue, 15 May 2007 11:25:42 +0800
>> From: Feizhou <feizhou at graffiti.net>
>> Subject: Re: [CentOS] Re: Anaconda doesn't support raid10
>> Message-ID: <46492836.3010808 at graffiti.net>
> 
> <snip>
> <the reliability factor has been proven to be the same across the board.
> <You must have missed the thread on the Google report on various drives
> <that they use
> 
> Feizhou:  Yes, I did not see that. If the MTBF's are the same, then the
> Performance specs you provided are what one needs to go with, when choosing
> components. If Google has posted Failure rates, for the drives they use, that 
> would be more meaningful than MTBF by the drive manufacturer, for that
> particular drive model.

They noted that there was no particular difference in failure rates 
whether they are high performance industrial drives or consumer drives. 
Basically they said that manufacturers claims of MBTF amounted to 
naught. It was the operating environment that determined the failure 
rate of drives, namely, fluctuating temperatures.

> 
> Questions: Is it your belief that all PATA/SATA/SCSI drives made by one
> manufacturer have the same MTBF? (I find that hard to believe. However,
> that might be the case.....) Also, is it your belief that these drives,
> made by different manufacturers, all have the same MTBF?

Bad batches aside, no I do not believe that different drives from 
different manufacturers have the same MTBF. Case in point are the IBM 
'glass' Deskstars. However, I personally choose Seagate drives and I 
have not had problems with them. In my previous job at an ISP, they used 
maxtors and western digital drives and I did not notice any particular 
brand being more troublesome (firmware and controller incompatibilities 
excepted) nor do I get the impression that they were better or worse 
than Seagate. I have no comment on newer manufacturers like Hitachi.

Due to the fact that those boxes run 24/7, I believe that that enabled 
consumer drives to last longer than they would due to less temperature 
fluctuations.



More information about the CentOS mailing list