All of lore.kernel.org
 help / color / mirror / Atom feed
From: Matthias Ferdinand <bcache@mfedv.net>
To: linux-bcache@vger.kernel.org
Subject: bad bcache performance - crappy SSD?
Date: Tue, 27 May 2014 00:16:49 +0200	[thread overview]
Message-ID: <20140526221649.GI10209@teapot> (raw)

Hi,

so far I have been unable to get an actual performance improvement out
of bcache, and I suspect that my SSD is just crap. Any help or hint
appreciated. What (SATA) SSDs have you been using successfully with
bcache?


My host system: Ubuntu 14.04 x64.
kernels tried: Ubuntu stock (3.13.0-24), 3.14, 3.15-rc{5,6}.

storage structure:
3xSATA (sda,sdb,sdc) => md raid5 (md125) => bcache0 => LVM (dm-1)

benchmark: install linux kernel header package inside kvm guest

        cache device not attached:            0:18 minutes
    SSD cache device attached (writethrough): 3:27 minutes
    SSD cache device attached (writeback):    3:25 minutes

With attached cache device, the ssd seems very busy (iostat %util),
while the raid members are not:

    Device:         rrqm/s   wrqm/s     r/s     w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util
(5) sdc               0.00    12.33    0.00   57.33     0.00    57.00     1.99     0.06    1.07    0.00    1.07   0.79   4.53
(5) sdb               0.33    12.00    0.33   57.00     2.67    54.33     1.99     0.07    1.19    0.00    1.19   0.84   4.80
(5) sda               0.00    11.33    0.00   57.00     0.00    51.67     1.81     0.06    1.10    0.00    1.10   0.80   4.53
(*) sdf               0.00     0.00    0.00   54.67     0.00    27.17     0.99     0.98   18.05    0.00   18.05  17.95  98.13

    (5) raid 5 member;  (*) SSD


when I replace the SSD with a single SATA disk:
    SATA cache device attached (writethrough): 0:19 minutes
    SATA cache device attached (writeback):    0:13 minutes


The SSD is an Intel 530, has current firmware and plenty of DISCARDed
blocks.  Specs claim >= 13.000 incompressible writes/sec for the 120GB
model, but the above data suggests it can handle just about 60.


    === START OF INFORMATION SECTION ===
    Device Model:     INTEL SSDSC2BW120A4
    Serial Number:    CVDA3393028N1207GN
    LU WWN Device Id: 5 5cd2e4 04b82aa5a
    Firmware Version: DC32
    User Capacity:    120,034,123,776 bytes [120 GB]
    Sector Size:      512 bytes logical/physical
    Rotation Rate:    Solid State Device
    Device is:        Not in smartctl database [for details use: -P showall]
    ATA Version is:   ACS-2 (minor revision not indicated)
    SATA Version is:  SATA 3.0, 6.0 Gb/s (current: 3.0 Gb/s)
    Local Time is:    Sat May 24 20:37:14 2014 CEST
    SMART support is: Available - device has SMART capability.
    SMART support is: Enabled



Regards
Matthias

             reply	other threads:[~2014-05-26 22:16 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-26 22:16 Matthias Ferdinand [this message]
2014-05-27 20:13 ` bad bcache performance - crappy SSD? Peter Kieser

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20140526221649.GI10209@teapot \
    --to=bcache@mfedv.net \
    --cc=linux-bcache@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.