Frank said:
Excuse me, Paul, but there is one thing I don't like: "You don't need
to know", since "I like to know!"
Let me explain: Raid 1 is no substitute for a backup, that's correct.
But if I receive RAID for free (with a P35 mainboard) and considering
the cost for a second hard drive (70 Euros for 0.5GB), it's simply
worth the money.
What I mean is: I pay 70 Euros for the second drive and I receive the
assurance "if a disk crashes (and only in that case) you'll have a
second one with the identical content". Nothing else! No virus
protection, no power problem protection, no whatsoever protection,
only "if one drive fails, the other has the copy and you don't need to
spend money, time and nerves to reconstruct from backup." Of course,
if the system blows up for any other reason, it's very helpful to have
a backup.
On the other hand, my personal statistic:I'm in that business for
nearly twenty years, I've started with a 20 MB MFM-harddrive (yes,
kids, it weight a few kilos!). In that time, I've experienced no virus
problem, no power problems, nothing, but two hard drives crashed in
only 27 months (1st an IBM and 2nd a Samsung). This is my personell,
not representative experience.
And after all these words: Does anybody know, what ICH9R does
(following the discussion of this thread)?
Kind regards
The metadata position is not public knowledge. If you want to know,
you connect a disk to an Intel Southbridge, create a RAID volume, and
then search for the sector or sectors that have changed.
There is a reasonable chance, that if you took a hard drive connected
as RAID1, on a ICH6R, it would work on a ICH9R. There should be continuity
in the forward direction. There might not be in the reverse direction
(as ICH9R might support RAID5, and ICH6R doesn't, and ICH6R RAID BIOS
won't know what RAID5 is).
The only other company, that offers some kind of statement about
compatibility between controllers, was Promise. Whether that is still
true, is unknown to me.
The reason no public statements are made, is it allows the companies
involved, to change the metadata format, any time they choose, without
warning to the public. As technical requirements merit.
As an example of my own, personal experience. I connected a drive
to a Promise controller (perhaps PDC20378), and discovered that the
first partition was no longer visible. So at least in the case
of that chip, and the metadata it was using, there was some problem
seeing a partition, when moving between an Intel chip and a separate
Promise PDC20378. Tomshardware did an article, some time ago,
where they tried something similar. I had my own theories as to
which test cases would work, and which ones would not. Virtually
any heterogenous case, failed. Some of the homogenous cases
(moving Promise array to another Promise card), worked.
But I don't see any categorical statements being possible, unless
these companies take inter working as a serious technical requirement.
You can buy two PCI or PCI Express controller cards, and put one away
for the day when the first one fails. As some of the cards are dirt
cheap (such as some of the old SIL3112 dual port cards), that is not
such an expensive alternative. But the performance and reliability
of these solutions, leaves a lot to be desired. For example, at
least one person has seen a SIL3112 operating in RAID1 mode,
where the RAID stopped "mirroring" for a period of three months.
When one of the disks died, the data available on the other disk
was "stale" and as near as the poster could determine, about
three months old. Based on that kind of performance, I'd want a
solution that is a bit better behaved.
I'd feel much better about motherboard RAID1 solutions, if I
knew the controller firmware or software, was auditing how
identical the disks were, during idle moments. I'm not aware
of any of these solutions, auditing their own performance.
Based on a few factors like that, I feel a backup is still
a wise investment.
Paul
