[xubuntu-users] Thoughts about SSD - Was: 18.04.1 Point Release Testing

[Ralf Mardorf]

On Sat, 2018-08-04 at 13:03 -0400, Art wrote:
> Encryption we have today, is just the home folder. If encryption of the
> entire drive is used, it doesn't take a rocket scientist to know that
> the performance slows severely, that more ram is needed and that more
> SSD read/write cycles will be needed. This increases the wear and tear
> on SSD's, which have a finite lifetime AND more system files need to be
> in ram to help avoid excessive delays.

Hi,

sorry, I couldn't resist to cross-post. A few days ago I mounted the
third SSD.

Take a look at the "cable clutter license" picture at 
https://i.imgur.com/DxcfbSz.png , it does show my PC's old and cheap
case. If HDDs are mounted to this case, you not only hear the even
silent HDDs, sometimes also the sidewalls of the case start vibrating
and make a very annoying noise.

I decided to treat my SSDs in the same way, I treated HDDs.

There are two main differences:

- When I used HDDs my PC usually was up 24/7, since I use SSDs, I tend
  to shutdown and to startup the PC each day.

- The SSDs gets trimmed once a week.

The used SSDs are more or less the cheapest available SSDs.
The vendor's Linux tool is very good, running

   gksudo ocz-ssd-utility &

it's easy to update the firmware, to get some special information and to
get much better SMART output, than provided by smartctl. Checking that
everything regarding the trimming is ok could be done by

   systemctl status fstrim.timer
   systemctl status fstrim.service
   journalctl -u fstrim.service

A 1,5 years old      TOSHIBA TL100 223.57 GiB did  cost 77,95 €.
A 1   year  old      TOSHIBA TL100 223.57 GiB did  cost 86,50 €.
A brand spanking new TOSHIBA TR200 223.57 GiB does cost 49,99 €.

A guess regarding the expected lifespan, based upon the health status
reported by the ocz-ssd-utility.

For the brand spanking new it's: 100%
For the 1   year  old it's:       88%
For the 1,5 years old it's:       65%

The expected lifespan might be around >= 3 years. The lifespan of my
last HDDs for the same kind of usage was around >= 7 years.

An unimportant drawback is usage with "vintage" installs.

My everyday Linux is an Arch Linux install. I'm using syslinux, not
grub, so to avoid chainloading, the Arch Linux install contains the
kernels of all other installs, too. Apart from Arch Linux, lets take a
look at two other Linux installs. The current up-to-date Ubuntu (server
image) 16.04.5 LTS install and the "vintage" Ubuntu 12.10 (Xubuntu or
Ubuntu Studio) install. All three installs support ext4 and are
installed on ext4 partitions.

Arch Linux was "cp -ai from_the_oldest_TL100 to_the_new_TR200", I
removed the boot flag from the original Arch Linux partition and added
it to the new Arch Linux copy and run "extlinux --install ...", to
install the bootloader. The two Ubuntu installs remain on a TL100 drive.

[rocketmouse at archlinux ~]$ grep PRETTY    /mnt/moonstudio/etc/os-release
PRETTY_NAME="Ubuntu 16.04.5 LTS"
[rocketmouse at archlinux ~]$ grep archlinux /mnt/moonstudio/etc/fstab
#dev/sda9                                   /mnt/archlinux ext4   defaults,relatime 0 2
/dev/sdc1                                   /mnt/archlinux ext4   defaults,relatime 0 2
/mnt/archlinux/.boot/ubuntu_moonstudio/boot /boot          none   bind              0 0

Remember, the Ubuntu kernels are on the Arch Linux partition. To mount
Arch Linux and to bind to /boot makes sense, to be able to upgrade a
kernel. This works without issues for Ubuntu 16.04.5 LTS and IIRC it
also worked without issues for the "vintage" Ubuntu 12.10 (EOL May 16,
2014), when the Arch Linux install was on the oldest TL100. After the
migration of the Arch Linux install to the new SSD it still works
for Ubuntu 16.04.5 LTS, but fails for the "vintage" Ubuntu 12.10. I get
a message that there should be a serious damage on the Arch Linux file
system, that can't be repaired, a newer version of ext4 is required or
fsck.ext4 or whatsoever. I don't care, I simply commented mounting
archlinux and to bind to /boot out in the "vintage" Ubuntu 12.10's
fstab.

Résumé

If the SSDs will last for >= 3 years it's ok for me. Since I didn't do
any troubleshooting regarding the issue with the "vintage" install, you
could assume that I don't care much about it. Even if it's said that due
to a missing cache those SSDs should be more or less the slowest SSDs on
the market they are unbelievable fast, absolutely fast enough regarding
my needs. To claim that they are slow is like claiming that a 290 km/h
fast car is disgusting slow, since all other cars of this kind are
usually 355 km/h fast. The fuel consumption might be much more
important. All SSDs need less current, than a HDD do, this could make a
big difference, especially for portable computers. I don't know if a so
called "slow" SSD in the end needs less current, than a "fast" SSD, but
I don't care, if a SSD is fast or faster, since I unlikely would notice
the difference.

Since I have long time experiences using HHDs I notice first signs if a
HDD starts to reach end of life. I don't know if there are any signs at
all, before a SSD dies. If a HDD is dead and a backup should miss a
little bit of data, it's usually possible to revitalize the HDD to
rescue the data. Would this be possible for a dead SSD, too?

What are your opinions, experiences and thoughts?

Regards,
Ralf