[Tfug] SSD plans & disk usage

[Louis Taber]

Hi All,

"Well, what are your workloads?"  from Tom Rini is an important question.
With a typically silly answer.
My "workloads" are usually:

0) X-window system
1) Firefox - with an occasional video.  (12 tabs currently)
2) Terminal sessions 10-15 at a time are not unusual.  Most local ones.
3) Some GIMP
4) Some Audacity  (I would like to do some 4 channel playback too. [latter])
5) rsync to remote sites for backup and such.

I did find an answer to one of my earlier questions -- what is the system
doing with each partition.

Take a look at:
http://linuxpoison.blogspot.com/2009/02/how-to-measure-and-read-disk-activity.html

In summary:
Field  1 -- # of reads completed successfully.
Field  2 -- # of reads merged,
Field  3 -- # of sectors read successfully.
Field  4 -- # of milliseconds spent reading
       (as measured from __make_request() to end_that_request_last()).
Field  5 -- # of writes writes completed successfully.
Field  6 -- # of writes merged
Field  7 -- # of sectors written successfully.
Field  8 -- # of milliseconds spent writing
        (as measured from __make_request() to end_that_request_last()).
Field  9 -- # of I/Os currently in progress --  should go to zero.
Field 10 -- # of milliseconds spent doing I/Os -- This field is increases so
long as field 9 is nonzero.
Field 11 -- weighted # of milliseconds spent doing I/Os

For my current system at the current time:
       top - 22:56:14 up 73 days, 14:50, 10 users,  load average: 0.37,
0.21, 0.12
                                          Sectors
read                                          Sectors written
Field#          1              2                  3
4              5              6                   7                 8
9            10              11
/            88436    246489      2678385    710744       7824      409043
  3335160      1796452    0    446712     2507092
/home  883287    262225 135901996  9254264  5200553    8927714 113119328
223525832   0  9396364 232783172
/tmp      45279      22991      5896086    486364    226505    2806769
24266992    19119160    0  1236676   19607592
/usr     746320    102178    11810500   7582124   376402    1428796
14476864    22569504    0  3041040   30285364
/var       91663      14062      1938596     620240   279402    1491145
14182000       6994996    0  1093092    7615088
/home/ltaber/alt-disk
             43054      10753      4086498    1206724   186631 16174886
130917320 226015104     0  4461956 227441464
*  Several partitions not listed.
(I moved a lot of data out of /home to /home/ltaber/alt-disk during this
time because of space issues.)

Or .48% of the weighted milliseconds doing I/O in / and /usr.  Probably not
worth putting the SSD at all.
And .08 is the total time (73days) divided by the weighted time doing I/O
--- what ever that is worth.

  - Louis


On Mon, Jan 17, 2011 at 9:42 PM, Tom Rini <trini at kernel.crashing.org> wrote:

>
> Well, what are your workloads?  iirc the big performance win from
> executions is when they've got a lot of data files to read (ie games),
> or if you're going to fire up gimp on stuff that's on SSD.  They're
> also quite good at speeding up your life if you're going to be
> reading/writing to them a lot (I know folks that use them for build
> space at work where there's an understanding that yes, the drive will
> die at some point sooner than expected, and that's OK).
>
> I guess what I'm driving at is that good SSDs can make use of the
> speeds SATA provides but if you aren't doing all that much or all that
> much that's intensive it's not going to be a win there.  Of course, if
> you're talking about a laptop, the no moving parts and all that is a
> win for its own reasons :)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tfug.org/pipermail/tfug_tfug.org/attachments/20110117/fd32b09f/attachment-0002.html>