Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> espoused:
> ____/ BearItAll on Wednesday 27 June 2007 11:38 : \____
>
>> Roy Schestowitz wrote:
>>
>>> Kernel space: two new filesystems for Linux
>>>
>>> ,----[ Quote ]
>>> | New filesystem technology for Linux includes high capacity,
>>> | snapshots, copy-on-write, and on-the-fly corruption detection.
>>> `----
>>>
>>> http://apcmag.com/6468/zfs_the_ultimate_filesystem
>>>
>>
>> Excuse the sweaty brow, I just feignted.
>>
>> Looking that up on the net it seems that on any error (yes I said 'any') the
>> system admin is notified and all further access to that system is stopped.
>>
>> Sorry, you just feignted too eh? Here, borrow my damp cloth.
>>
>> So, you get an email from your system 500 miles away, it so happens to be
>> the day when the ISP is doing some scheduled work so the email is a little
>> later than normal. Telling you that the file system has a little error and
>> has therefore stopped.
>>
>> Thankyou for the offer of a new file system, but I don't think it's the
>> right one for me.
>>
>> Ah but, I here you say, assuming that is a raided drive then whats the
>> problem, the bad side of the raid is kicked out, but the system still runs.
>>
>> Well pardon me but I don't want to have to dash off 400 miles (even a
>> virtual dash through vpn) every time we have a bad sector. What do these
>> people think goes on currently? Linux is better than that, we don't need to
>> kill off a drive or volume for a large proportion of potential drive
>> errors. Of those problems that are marked for the filesystem check over
>> night, how many would actually bring down the system? The answer is, very
>> few because the system is well able to deal with these situations these
>> days.
>>
>> There, I've spat my dummy and quite right too.
>
> I think that Vista devs tried to brag about on-the-fly(ish) defragmentation of
> the disk (allocation), but again, the problem here is that a problem is being
> solved using CPU cycles rather then be prevented in an elegant way in the
> first place.
>
I can see that there could be some applications where this behaviour
might be exactly what you wanted, though. You could get /very/ cute
and RAID together several instances, say, with the certain knowledge
that if one of the filesystems went out of kilter, then it would be
withdrawn from the array instantly. So long as you're using the right
kind of array, that could be useful. Another example might be for,
say, critical records, medical ones, or perhaps a billing system where
you hvae, say, 20-30 Million bills/quarter to produce, for example.
Filesystem errors for such a system could be seen as being truly critical.
--
| Mark Kent -- mark at ellandroad dot demon dot co dot uk |
| Cola faq: http://www.faqs.org/faqs/linux/advocacy/faq-and-primer/ |
| Cola trolls: http://colatrolls.blogspot.com/ |
| My (new) blog: http://www.thereisnomagic.org |
|
|
