On Mar 9, 2011, at 7:48 PM, Dan Shoop wrote: > > On Mar 9, 2011, at 4:14 PM, Chris Murphy wrote: >> In any event, that one is expected to take precautions with RAID 5/6 or even >> RAID-Z with respect to power management does not mean the incidence of >> silent data corruption is higher. > > RAID-Z, raids 0, 1, 10 0+1 are not capable of silent data corruption. All > corruption is checked and managed due to the nature of the operations of > these configurations.
Silent data corruption can occur on all of those RAID versions. That there is eventually error detection and potentially error correction does not mean the silent data corruption did not occur. The incident of it happening in the first place, and the consequence of that silent data corruption to the end user's data are two different things. It's imply wrong to suggest, as you are that a.) there is a higher incident of silent data corruption with respect to RAID 5 and that the above listed RAID are not "capable" of silent data corruption. A bad capable can absolutely render the data on RAID 1 corrupt and you can have data loss. Same for RAID 0, 10, 0+1. RAID-Z is also not immune to the original instance of such corruption, it simply recovers from it better. > >> It simply means *IF* it happens to a RAID 5 array the problems can rapidly >> become magnified requiring significant contingencies. > > No it means that the others can't have it but RAID5 can, and will. That's ridiculous. And unsupportable rubbish. "Silent data corruption" != RAID 5 write hole/parity pollution. And the write hole does not require *any* user data become corrupted at all. The write hole is explicitly corruption of RAID parity data. No data loss occurs unless there are additional failures like blocks or drives requiring that parity data to be used for restoration - if the parity data needed is corrupt, the original data cannot be restored. You can have silent data corruption of file system structures, metadata, or file data itself. That has nothing to do with RAID 5 and it absolutely can affect RAID 1 and non-arrayed volumes. > And to a degree that previously wasn't seen until very large numbers of raid5 > arrays were seen in use. Now that the industry has seen lots of raid5 use, > and big big farms of computers that had been using raid5 started to notice an > unexpected prevalence of silent data corruption, and disks have gotten > cheaper making raid5 an unnecessary risk, many wise men have decided it's not > being diligent to use raid5. There's no question there are better solutions for the detection and correction of corruption. > >> I might still be able to suck off a bunch of data from a non-array (or RAID >> 1 disk) despite file system corruption, or the corruption of even 10% of the >> disk - I will have *better* recovery from RAID 5 if that same event happens >> to 1 disk, but much worse recovery if that problem is propagated through the >> entire array. I don't think this is a secret. Again, RAID 5 is not a new >> thing. > > You really don't understand silent data corruption. You are clearly demonstrating to the list you do not understand it. You are making it out to be the same thing as the RAID 5 write hole. And you're saying completely unsupportable things like RAID 1 is simply not capable of silent data corruption. But you are invited to explain how RAID 1 is immune to silent data corruption. > > Data on a non-raided or raid 0 disk, it's silent data damage, and tracked by > the bad block replacement. Silent data damage? That is a fantasy term. You're going to state that silent data corruption is unique to RAID 4 and above? Really? > >> But I totally disagree with the wording used that implies there is a high >> incidence of silent data corruption inherent to a RAID 5 system. > > Well since other disk subsystem architectures can't have it, that's about as > high of a difference as you can get, measurably. There are tens of thousands of web site, white papers, research papers that do not at all tie the term "silent data corruption" to RAID 4 and above. You are making out parity pollution to be the same thing as silent data corruption. Parity pollution is a kind of silent data corruption in certain cases (arguably not when it's well known to be possible on a power failure or crashed system). Parity pollution is not the only kind of silent data corruption. It is not unique to RAID 4 and above. Chris Murphy_______________________________________________ MacOSX-admin mailing list [email protected] http://www.omnigroup.com/mailman/listinfo/macosx-admin
