Re: Progress on my new Debian box
Bzzzz <lazyvirus@gmx.com> wrote:>
> On Mon, 23 Jun 2014 12:43:19 -0400
> Steve Litt <slitt@troubleshooters.com> wrote:
>> I've got the hardware all set up. AMD dual core 4100, 16GB RAM, 240GB
>> SSD for /, 750GB Western Digital Black for /var, /tmp, /run, and swap
>> partition,
> As a SSD has limited write capacities, people usually avoid using it
> for things that are often (re)written. Unfortunately, you just
> indicate _all_ wrong directories to store on a SSD…
I beg to differ. (I know, you already commented on your error while
reading Steves sentence.)
Modern SSDs are no more fragile than a normal 2,5" spinning drive or
even less, since they have no problems with vibration or sudden G
shocks.
Have a look at
http://techreport.com/review/26523/the-ssd-endurance-experiment-casualties-on-the-way-to-a-petabyte
,----
| We started with six SSDs: the Corsair Neutron GTX 240GB, Intel 335
| Series 240GB, Samsung 840 Series 250GB, Samsung 840 Pro 256GB, and two
| Kingston HyperX 3K 240GB.
| [...]
| The last time we checked in, the SSDs had just passed the 600TB mark
| [URL1]. They were all functional, but the 840 Series was burning
| through its TLC cells at a steady pace, and even some of the MLC drives
| were starting to show cracks. We've now written over a petabyte, and
| only half of the SSDs remain. Three drives failed at different
| points—and in different ways—before reaching the 1PB milestone.
`----
URL1 http://techreport.com/review/26058/the-ssd-endurance-experiment-data-retention-after-600tb
Of course, if you heavily abuse your SSD (like the techreport guys are
doing it), you will have a bad time.
But you can just use the drive like a normal hard drive, no need to
specially protect them like a raw egg, aside from leaving a bit of space
(10%) unused so the controller has more room for its wear leveling
algorithm.
My own data points for the Samsung 840 256GB Evo in my laptop look like this.
5505 hours running (~230 days) with ~3450GiB written. Wear_Leveling_Count
is at 98, Reallocated_Sector_Ct is at 100.
,----
| ID# ATTRIBUTE_NAME VALUE WORST THRESH TYPE RAW_VALUE
| 5 Reallocated_Sector_Ct 100 100 010 Pre-fail 0
| 9 Power_On_Hours 098 098 000 Old_age 5505
| 12 Power_Cycle_Count 099 099 000 Old_age 430
| 177 Wear_Leveling_Count 098 098 000 Pre-fail 18
| 179 Used_Rsvd_Blk_Cnt_Tot 100 100 010 Pre-fail 0
| 181 Program_Fail_Cnt_Total 100 100 010 Old_age 0
| 182 Erase_Fail_Count_Total 100 100 010 Old_age 0
| 183 Runtime_Bad_Block 100 100 010 Pre-fail 0
| 187 Uncorrectable_Error_Cnt 100 100 000 Old_age 0
| 190 Airflow_Temperature_Cel 065 052 000 Old_age 35
| 195 ECC_Error_Rate 200 200 000 Old_age 0
| 199 CRC_Error_Count 100 100 000 Old_age 0
| 235 POR_Recovery_Count 099 099 000 Old_age 12
| 241 Total_LBAs_Written 099 099 000 Old_age 7233847674
`----
There is the Swap space, /var and /tmp on that SSD and I suspend-to-disk
the laptop every morning and evening.
I still expect this drive to outlive the laptop it is currently inside
by far.
You also can quite precisely pinpoint the moment a SSD will fail and
backup your data before while a spinning harddrive may fail without
warning at any moment.
I would advice Steve to put Swap and /var back onto the SSD as those are
the areas which gain the most from being on a fast storage medium.
/tmp can be a tmpfs, which will be written to the Swap if needed, also
profiting from the fast SSD.
/run should already be a tmpfs in Debian Wheezy.
/home can then be on the WD because of the big space available there.
Grüße,
Sven.
--
Sigmentation fault. Core dumped.
Reply to: