[svlug] SSD and workstation

[Jesse Monroy]

Greg,
thanks for jumping in. I'm having a hard time believing that
you are experiencing wear issues. I've heard from many, many
sysadmins on many related issues; most notably the read/write
delay encountered overtime. I'm sure I heard you say similar.

As for wear, I am taking on the word of Admins and vendors.
Perhaps you can clarify why bleeko is NOT encountering
these issues. Are you using SSD as "read only"? Or similar?

Jesse


On 12/17/12, Greg Lindahl <lindahl at pbm.com> wrote:
> On Sun, Dec 16, 2012 at 11:12:12PM -0800, Jesse Monroy wrote:
>> On 12/16/12, Marco Walther <marco at sonic.net> wrote:
>> > I don't know if that's true any more (or if it ever was). I don't think
>> > you would wear out a even a `consumer grade' SSD in a reasonable time
>> > frame. noatime is probably a good idea (for any workstation/laptop/...
>> > were you don't really are about the access times) but otherwise I would
>> > use the SSD wre you expect the most accesses.
>> >
>> You don't need to take my word for it. Ask Admins that
>> work for Yahoo, LinkedIn, Google, or http://blekko.com/
>
> Or blekko might even proactively answer you  :-)
>
> We have about 55 systems which are running Linux with SSD /, they do
> not have noatime set, and, we haven't had any wearout issue. These are
> MLC drives, Intel X-25Ms, so we'd notice if there was a
> problem. Here's my desktop:
>
> [root at greg-desk b]$ /usr/sbin/smartctl -a /dev/sda
> ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED
> WHEN_FAILED RAW_VALUE
>   9 Power_On_Hours          0x0032   100   100   000    Old_age   Always
>   -       23840
> 232 Available_Reservd_Space 0x0033   099   099   010    Pre-fail  Always
>   -       0
> 233 Media_Wearout_Indicator 0x0032   097   097   000    Old_age   Always
>   -       0
>
> In this case the OP was looking at a 32 gbyte drive, so it's cheap to
> replace if it turns out to be a problem.
>
> -- greg (blekko cto)
>
>
>