This is a little bit off-topic, but still about IT. :-)
SSD and RAID and device failure.
1. RAID6. n+2 means MTTDL (Mean Time To Data Lost) is significantly
lower than RAID5 or RAID10.
2. SSD module itself. It is no longer (or have never been) simple chip
from pendrive or even SSD from laptop. Current SSD modules are computers
with complex algorithms for monitoring and moving logical hotspots over
the physical SSD "area". And they have redundant area. A lot of -
depending on SSD class.
3. Even shiny new RADI6 SSD array may fail. This is the case when you
lose your logical tape volumes on it. ...unless you have DR solution,
that means another VTS with new shiny RAID6 and copy of your volumes.
All only those you decided to replicate.
4. Did I say second VTS? You may have yet another. And another.
Geographically dispersed.
5. Don't forget to pay for the above. :-) Seriously: are your
requirements justified and backed up by your budget?
--
Radoslaw Skorupka
Lodz, Poland
W dniu 17.09.2024 o 22:40, Mike Schwab pisze:
Also, if using SSD in a raid array, you probably won't get a warning
of impending failure, and you usually have devices with very similar
lifetime of write and write history. Which can lead to the entire
raid group going at once.
https://www.reddit.com/r/homelab/comments/17h1xgh/how_do_you_mitigate_the_risk_of_simultaneous/
On Tue, Sep 17, 2024 at 3:13 PM Radoslaw Skorupka
<[email protected]> wrote:
W dniu 17.09.2024 o 18:04, Steve Estle pisze:
A healthy tape would be defiined as a tape where you can successfully read the
tape label and all written blocks on the tape volser without error (I/O error,
equipment check, etc).
Well, it is not so simple.
1. Media. Healthy tape media is no longer important for majority of
installations, because nowadays we have VTS (any vendor) and last
physical tape (TS1140) is really old, no longer supported, etc. However
yes, healthy physical tape media means the tape can be read. From
beginning to end of volume.
2. Healthy tape format. Yes, tapes have format. Labels, etc. Healthy
tape means existing labels and consistence between labels and between
labels and datasets (i.e. number of blocks).
3. Healthy virtual tape "media". Actually the tape volume is a file. The
file is usually OK, however one can imagine broken format, i.e.
accidentally overwritten fileds or other logical errors. Very unlikely
IMHO, however some malware on VTS or just errors in microcode could do
this.
4. Healthy physical media in VTS. Usually a VTS cache is RAID-protected
disk array, so the problems seems to be unlikely. Of course some VTSes
can offload virtual tape volumes on real tape - in that case the tape
can be duplicated or not. However such error could mean lost of volume
at all.
My €0.02
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
