[opensuse-factory] Improving IO with Sata HDD
Dears Tumbleweed users, gurus and hackers, I would like to get some suggestions by more skilled people in order to improve my low IO disk performance because sometimes by using iotop I can see over 99% of disk activity especially when running VirtualBox and packagekit is updating the system. HDD activity led is always on during high IO activity and system sometimes get unresponsive or in the better case very very slow. I tried to change one ext4 parameter from [cfq] to [noop] without any noticeable improvements. sudo echo noop > /sys/block/sda/queue/scheduler Would like to know if I can try some other tuning. My HDD is an ATA 465,8G SAMSUNG HM501II ext4 Many thanks and regards! Cheers, -- Marco Calistri Opensuse Tumbleweed 64 bit - Kernel 4.7.0-2-default Gnome 3.20 Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 - Intel® Sandybridge Mobile
Hi Marco, On Fri, Dec 2, 2016 at 2:05 PM, Marco Calistri <mcalistri@hotmail.com> wrote:
Dears Tumbleweed users, gurus and hackers,
I would like to get some suggestions by more skilled people in order to improve my low IO disk performance because sometimes by using iotop I can see over 99% of disk activity especially when running VirtualBox and packagekit is updating the system.
HDD activity led is always on during high IO activity and system sometimes get unresponsive or in the better case very very slow.
I tried to change one ext4 parameter from [cfq] to [noop] without any noticeable improvements.
sudo echo noop > /sys/block/sda/queue/scheduler
ISTR that noop would be beneficial for SSDs, but not for HDDs. Other than that, I suggest that you use general Linux tuning guides available online, I don't think there are any specific openSUSE tweaks. Best, Robert
Would like to know if I can try some other tuning.
My HDD is an ATA 465,8G SAMSUNG HM501II ext4
Many thanks and regards!
Cheers,
-- Marco Calistri Opensuse Tumbleweed 64 bit - Kernel 4.7.0-2-default Gnome 3.20 Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 - Intel® Sandybridge Mobile
-- http://robert.muntea.nu/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 04/12/2016 06:25, Robert Munteanu ha scritto:
Hi Marco,
Dears Tumbleweed users, gurus and hackers,
I would like to get some suggestions by more skilled people in order to improve my low IO disk performance because sometimes by using iotop I can see over 99% of disk activity especially when running VirtualBox and packagekit is updating the system.
HDD activity led is always on during high IO activity and system sometimes get unresponsive or in the better case very very slow.
I tried to change one ext4 parameter from [cfq] to [noop] without any noticeable improvements.
sudo echo noop > /sys/block/sda/queue/scheduler ISTR that noop would be beneficial for SSDs, but not for HDDs. Other
On Fri, Dec 2, 2016 at 2:05 PM, Marco Calistri <mcalistri@hotmail.com> wrote: than that, I suggest that you use general Linux tuning guides available online, I don't think there are any specific openSUSE tweaks.
Best,
Robert
Would like to know if I can try some other tuning.
My HDD is an ATA 465,8G SAMSUNG HM501II ext4
Many thanks and regards!
Cheers,
-- Marco Calistri Opensuse Tumbleweed 64 bit - Kernel 4.7.0-2-default Gnome 3.20 Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 - Intel® Sandybridge Mobile
Thanks for your feedback Robert! I rolled back to [cfq] now but want to try the parameter vm.swappiness which as default is =60 and set it to =10: cat /etc/sysctl.d/99-sysctl.conf vm.swappiness=10 Cheers, Marco N�����r��y隊Z)z{.���r�+�맲��r��z�^�ˬz��N�(�֜��^� ޭ隊Z)z{.���r�+��0�����Ǩ�
On 12/04/2016 08:49 AM, Marco Calistri wrote:
Il 04/12/2016 06:25, Robert Munteanu ha scritto:
Hi Marco,
Thanks for your feedback Robert! I rolled back to [cfq] now but want to try the parameter vm.swappiness which as default is =60 and set it to =10: cat /etc/sysctl.d/99-sysctl.conf vm.swappiness=10 Cheers, Marco ------- Guys, Check out this guide for openSUSE users. http://bit.ly/2g3rhO1 Cheers! Roman -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sun, 2016-12-04 at 13:49 +0000, Marco Calistri wrote:
I rolled back to [cfq] now but want to try the parameter vm.swappiness which as default is =60 and set it to =10:
cat /etc/sysctl.d/99-sysctl.conf
vm.swappiness=10
Rather than playing with tuning parameters, I'd suggest to proceed more systematically. Run some benchmarks to find out the "raw" performance of your disk for sequential and random IO. Do all this with the "noop" scheduler. I usually start out with a plain "dd" sequential read from the block device with 1M block size. If you have an unused partition, try sequential writes as well. Compare that to the device specs (or possibly with measurements you find on the net). This will give you an idea what to expect from the HW, and whether the HW is operating in the expected speed range. Next, run some simple benchmarks like bonnie++ on your file system. Be sure not to measure the page cache (use direct IO). Try different IO schedulers to see if it makes a difference. Finally, under your problematic load (virtualbox + packagekit), run iostat, vmstat or something similar and compare the disk throughput numbers you obtain with those from the previous steps. If the numbers (roughly) match the values to be expected from the HW, you'll either have to look for faster storage HW, or you need to distribute your IO load better. Especially if you have lots of RAM a lot of writes, you may want to look into the vm.dirty* sysctl settings. My personal rule of thumb is "set dirty_bytes to the number of bytes that the IO HW can write in ~1 second, and dirty_background_bytes to 10x the previous value". Regards Martin
Cheers,
Marco Nry隊Z)z{.r+맲rz^ˬzN(֜^ ޭ隊Z)z{.r+0Ǩ
-- Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Just my 2¢: I probably should mention GNOME Disks application — it ships simple yet still illustrative benchmarking tool I find extremely useful when occasionally testing new hard drives. 2016-12-07 13:53 GMT+03:00 Martin Wilck <mwilck@suse.com>:
On Sun, 2016-12-04 at 13:49 +0000, Marco Calistri wrote:
I rolled back to [cfq] now but want to try the parameter vm.swappiness which as default is =60 and set it to =10:
cat /etc/sysctl.d/99-sysctl.conf
vm.swappiness=10
Rather than playing with tuning parameters, I'd suggest to proceed more systematically. Run some benchmarks to find out the "raw" performance of your disk for sequential and random IO. Do all this with the "noop" scheduler. I usually start out with a plain "dd" sequential read from the block device with 1M block size. If you have an unused partition, try sequential writes as well. Compare that to the device specs (or possibly with measurements you find on the net). This will give you an idea what to expect from the HW, and whether the HW is operating in the expected speed range.
Next, run some simple benchmarks like bonnie++ on your file system. Be sure not to measure the page cache (use direct IO). Try different IO schedulers to see if it makes a difference.
Finally, under your problematic load (virtualbox + packagekit), run iostat, vmstat or something similar and compare the disk throughput numbers you obtain with those from the previous steps. If the numbers (roughly) match the values to be expected from the HW, you'll either have to look for faster storage HW, or you need to distribute your IO load better.
Especially if you have lots of RAM a lot of writes, you may want to look into the vm.dirty* sysctl settings. My personal rule of thumb is "set dirty_bytes to the number of bytes that the IO HW can write in ~1 second, and dirty_background_bytes to 10x the previous value".
Regards Martin
Cheers,
Marco Nry隊Z)z{.r+맲rz^ˬzN(֜^ ޭ隊Z)z{.r+0Ǩ
-- Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg)
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-- Regards, Andrei Dziahel -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 2016-12-07 11:53, Martin Wilck wrote:
Rather than playing with tuning parameters, I'd suggest to proceed more systematically. Run some benchmarks to find out the "raw" performance of your disk for sequential and random IO. Do all this with the "noop" scheduler. I usually start out with a plain "dd" sequential read from the block device with 1M block size. If you have an unused partition, try sequential writes as well. Compare that to the device specs (or possibly with measurements you find on the net). This will give you an idea what to expect from the HW, and whether the HW is operating in the expected speed range.
You can measure the raw read speed with hdparm -Tt /dev/DEVICE -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)
Il 07/12/2016 13:12, Carlos E. R. ha scritto:
On 2016-12-07 11:53, Martin Wilck wrote:
Rather than playing with tuning parameters, I'd suggest to proceed more systematically. Run some benchmarks to find out the "raw" performance of your disk for sequential and random IO. Do all this with the "noop" scheduler. I usually start out with a plain "dd" sequential read from the block device with 1M block size. If you have an unused partition, try sequential writes as well. Compare that to the device specs (or possibly with measurements you find on the net). This will give you an idea what to expect from the HW, and whether the HW is operating in the expected speed range. You can measure the raw read speed with hdparm -Tt /dev/DEVICE
Thanks to all replying to my question. Carlos, here you can see the result of the command you suggested: marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root: /dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec But I'm currently using the bfq scheduler which after some days of usage I can definetly affirm performs better than cfq with my system. Regards, -- Marco Calistri Opensuse Tumbleweed 64 bit Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 Intel® Sandybridge Mobile
Il 09/12/2016 00:49, Marco Calistri ha scritto:
marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root:
/dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec
HD Model ? It seems slow.. Here with an old Seagate Barracuda: /dev/sda: Timing cached reads: 13828 MB in 2.00 seconds = 6920.66 MB/sec Timing buffered disk reads: 308 MB in 3.00 seconds = 102.53 MB/sec Bye, Daniele. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 08/12/2016 22:03, Daniele ha scritto:
Il 09/12/2016 00:49, Marco Calistri ha scritto:
marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root:
/dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec
HD Model ? It seems slow..
Here with an old Seagate Barracuda: /dev/sda: Timing cached reads: 13828 MB in 2.00 seconds = 6920.66 MB/sec Timing buffered disk reads: 308 MB in 3.00 seconds = 102.53 MB/sec
Bye, Daniele. Hi Daniele, It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
Cheers, -- Marco Calistri Opensuse Tumbleweed 64 bit Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 Intel® Sandybridge Mobile
Il 08/12/2016 22:09, Marco Calistri ha scritto:
Il 08/12/2016 22:03, Daniele ha scritto:
Il 09/12/2016 00:49, Marco Calistri ha scritto:
marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root:
/dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec HD Model ? It seems slow..
Here with an old Seagate Barracuda: /dev/sda: Timing cached reads: 13828 MB in 2.00 seconds = 6920.66 MB/sec Timing buffered disk reads: 308 MB in 3.00 seconds = 102.53 MB/sec
Bye, Daniele. Hi Daniele, It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
Cheers,
Forgot to say that I'm using ext4 here! -- Marco Calistri Opensuse Tumbleweed 64 bit Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 Intel® Sandybridge Mobile N�����r��y隊Z)z{.���r�+�맲��r��z�^�ˬz��N�(�֜��^� ޭ隊Z)z{.���r�+��0�����Ǩ�
Il 09/12/2016 01:13, Marco Calistri ha scritto:
It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
Cheers,
Forgot to say that I'm using ext4 here!
HDD is slow, nothing to do.. It's time to move on SSD or Hybrid if you want better performance. Daniele. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
------------------------------------------------------------------------ *From:* Daniele *Sent:* 08/12/2016 22:18:52 (-0200 UTC) *To:* Opensuse-factory *Subject:* Re: [opensuse-factory] Improving IO with Sata HDD
Il 09/12/2016 01:13, Marco Calistri ha scritto:
It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
Cheers,
Forgot to say that I'm using ext4 here!
HDD is slow, nothing to do.. It's time to move on SSD or Hybrid if you want better performance.
Daniele. Got it, Thanks! N�����r��y隊Z)z{.���r�+�맲��r��z�^�ˬz��N�(�֜��^� ޭ隊Z)z{.���r�+��0�����Ǩ�
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2016-12-09 01:09, Marco Calistri wrote:
Il 08/12/2016 22:03, Daniele ha scritto:
Il 09/12/2016 00:49, Marco Calistri ha scritto:
HD Model ? It seems slow..
Hi Daniele, It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
How many RPM? Maybe it is a 5200 RPM model. That speed would be typical of such. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhKDO8ACgkQja8UbcUWM1w9GwD+Jf5lDPIODBvSgfIoIUq1zenL 8pnjusbajqV0ni8OyScA/3eKb+NtIs2rwjfqlna0rTLR3qNnrDiyQTM3tvcKrNxI =47ng -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 9 December 2016 at 09:46, Carlos E. R. <robin.listas@telefonica.net> wrote:
How many RPM?
Maybe it is a 5200 RPM model. That speed would be typical of such.
Searching it up online, it seems to be a "5400 RPM 8MB Cache SATA 3.0Gb/s 2.5" Internal Notebook Hard Drive", which would (at least partially) explain things. - Karl Cheng (Qantas94Heavy) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
------------------------------------------------------------------------ *From:* Carlos E. R. *Sent:* 08/12/2016 23:46:23 (-0200 UTC) *To:* Opensuse-factory *Subject:* Re: [opensuse-factory] Improving IO with Sata HDD
On 2016-12-09 01:09, Marco Calistri wrote:
Il 08/12/2016 22:03, Daniele ha scritto:
Il 09/12/2016 00:49, Marco Calistri ha scritto:
HD Model ? It seems slow..
Hi Daniele, It is a SAMSUNG 500G, SAMSUNG HM501II ATA.
How many RPM?
Maybe it is a 5200 RPM model. That speed would be typical of such.
Yes of course is the slower speed, I would be very surprised it being the 7200 RPM model. Anyhow with bfq scheduler is accettable this is a notebook and not a multi-core 24x8 server ;-) Cheers, Marco N�����r��y隊Z)z{.���r�+�맲��r��z�^�ˬz��N�(�֜��^� ޭ隊Z)z{.���r�+��0�����Ǩ�
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2016-12-09 13:05, Marco Calistri wrote:
*From:* Carlos E. R.
How many RPM?
Maybe it is a 5200 RPM model. That speed would be typical of such.
Yes of course is the slower speed, I would be very surprised it being the 7200 RPM model. Anyhow with bfq scheduler is accettable this is a notebook and not a multi-core 24x8 server ;-)
Well, your disk has the normal speed for its class. You can use a disk with a faster rotating speed, which will transfer data faster, and also use more battery. Or switch to a flash disk (watch the sustained write speed), which is faster but has a shorter life and is more expensive for the same size. Uses less battery, I believe. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhKoA8ACgkQja8UbcUWM1zhtwD/S6N+ksxumlUf/Pvoo+hdqgUJ UxgE4WhM6dDTkzcnupIA/0dz4UOm5N3JL5Wr2z+uGW9EZAPSAYP8nABaIrHwO+UH =5J+I -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Friday 2016-12-09 13:05, Marco Calistri wrote:
How many RPM? Maybe it is a 5200 RPM model. That speed would be typical of such.
Yes of course is the slower speed, I would be very surprised it being the 7200 RPM model.
Rotational rate however has more impact on random reads (seeks) than transfer speed for streaming (linear) reads. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2016-12-09 13:21, Jan Engelhardt wrote:
On Friday 2016-12-09 13:05, Marco Calistri wrote:
How many RPM? Maybe it is a 5200 RPM model. That speed would be typical of such.
Yes of course is the slower speed, I would be very surprised it being the 7200 RPM model.
Rotational rate however has more impact on random reads (seeks) than transfer speed for streaming (linear) reads.
On both. hdparm -tT goes up to 100 MB/s at typically. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhKpYUACgkQja8UbcUWM1ztowD9FyH23s+wTo21EXgcrPMJWR4L vnNZ7HuOLh5U+8uJHsMA/3Ob/6kikzI8O5gO/KlHncGFAGiXWZpU7cWC6eCF+pNT =Bfyw -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
* Marco Calistri <mcalistri@hotmail.com> [12-08-16 18:52]:
[....]
Thanks to all replying to my question.
Carlos, here you can see the result of the command you suggested:
marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root:
/dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec
But I'm currently using the bfq scheduler which after some days of usage I can definetly affirm performs better than cfq with my system.
definitely slow however you have it configured. My Intel 128gb 4 year old: hdparm -Tt /dev/sdc /dev/sdc: Timing cached reads: 17750 MB in 2.00 seconds = 8881.27 MB/sec Timing buffered disk reads: 972 MB in 3.00 seconds = 323.57 MB/sec Tw with btrfs -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Photos: http://wahoo.no-ip.org/gallery2 Registered Linux User #207535 Photos: http://wahoo.no-ip.org/piwigo @ http://linuxcounter.net -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2016-12-09 00:49, Marco Calistri wrote:
Carlos, here you can see the result of the command you suggested:
marco@linux-turion64:~> sudo hdparm -Tt /dev/sda [sudo] password di root:
/dev/sda: Timing cached reads: 9168 MB in 2.00 seconds = 4585.57 MB/sec Timing buffered disk reads: 232 MB in 3.02 seconds = 76.71 MB/sec
An interesting exercise is to partition an empty disk in, say, 20 equally sized partitions, and measure the speed on each one. And do it for different filesystems, too. You probably will find out that the disk is faster about 1/3 of the way. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhKCuUACgkQja8UbcUWM1yVXgD+ILRbp33EXKCmHEt8JE1now+q wJ6GNueGZbQ9I/bs5YMA/jLGMJPF64B+d8qnNSKDCXMVUCR5A/X12gg6oG1LdzD2 =MjAb -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Fri, 2016-12-09 at 02:37 +0100, Carlos E. R. wrote:
An interesting exercise is to partition an empty disk in, say, 20 equally sized partitions, and measure the speed on each one. And do it for different filesystems, too.
You probably will find out that the disk is faster about 1/3 of the way.
True for sequential IO, for obvious reasons (more data on a single track in the outer part of the disk). For random IO, you'll have an average wait time of a half rotation, plus head movement, and both is independent on the length of the track. Martin -- Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2016-12-09 21:13, Martin Wilck wrote:
On Fri, 2016-12-09 at 02:37 +0100, Carlos E. R. wrote:
An interesting exercise is to partition an empty disk in, say, 20 equally sized partitions, and measure the speed on each one. And do it for different filesystems, too.
You probably will find out that the disk is faster about 1/3 of the way.
True for sequential IO, for obvious reasons (more data on a single track in the outer part of the disk). For random IO, you'll have an average wait time of a half rotation, plus head movement, and both is independent on the length of the track.
The curious thing is that outer part of the disk is often slower than some way into the disk. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhLH/IACgkQja8UbcUWM1zlTgD7BSKnLHs3gk0Ucg6MNXYsuqb2 F/iGgrnyDFEF2ijcwIsA/2o5epJ1xP+n9tIMVRVLXBny32VXoEXr1C8dTD4EKHMV =HFtn -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 02/12/2016 13:05, Marco Calistri ha scritto:
Dears Tumbleweed users, gurus and hackers,
I would like to get some suggestions by more skilled people in order to improve my low IO disk performance because sometimes by using iotop I can see over 99% of disk activity especially when running VirtualBox and packagekit is updating the system.
HDD activity led is always on during high IO activity and system sometimes get unresponsive or in the better case very very slow.
I tried to change one ext4 parameter from [cfq] to [noop] without any noticeable improvements.
sudo echo noop > /sys/block/sda/queue/scheduler
Would like to know if I can try some other tuning.
My HDD is an ATA 465,8G SAMSUNG HM501II ext4
Many thanks and regards!
Cheers,
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler. Bye, Daniele. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 04/12/2016 07:06, Daniele ha scritto:
Il 02/12/2016 13:05, Marco Calistri ha scritto:
Dears Tumbleweed users, gurus and hackers,
I would like to get some suggestions by more skilled people in order to improve my low IO disk performance because sometimes by using iotop I can see over 99% of disk activity especially when running VirtualBox and packagekit is updating the system.
HDD activity led is always on during high IO activity and system sometimes get unresponsive or in the better case very very slow.
I tried to change one ext4 parameter from [cfq] to [noop] without any noticeable improvements.
sudo echo noop > /sys/block/sda/queue/scheduler
Would like to know if I can try some other tuning.
My HDD is an ATA 465,8G SAMSUNG HM501II ext4
Many thanks and regards!
Cheers,
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler.
Bye, Daniele.
Hi Daniele, BFQ is not available on m TW so can't give it a try. marco@linux-turion64:~> cat /sys/block/sda/queue/scheduler noop deadline [cfq] Thanks and regards, -- Marco Calistri Opensuse Tumbleweed 64 bit Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 Intel® Sandybridge Mobile
Il 04/12/2016 14:53, Marco Calistri ha scritto:
Hi Daniele,
BFQ is not available on m TW so can't give it a try. marco@linux-turion64:~> cat /sys/block/sda/queue/scheduler noop deadline [cfq]
Thanks and regards,
Yes, not available (by default) on 13.2 too. Search on software.opensuse.org ! Bye, Daniele. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 04/12/2016 12:06, Daniele ha scritto:
Il 04/12/2016 14:53, Marco Calistri ha scritto:
Hi Daniele,
BFQ is not available on m TW so can't give it a try. marco@linux-turion64:~> cat /sys/block/sda/queue/scheduler noop deadline [cfq]
Thanks and regards,
Yes, not available (by default) on 13.2 too. Search on software.opensuse.org !
Bye, Daniele.
Ok! Thanks and regards, -- Marco Calistri Opensuse Tumbleweed 64 bit Intel® Core™ i5-2410M CPU @ 2.30GHz × 4 Intel® Sandybridge Mobile
On 12/04/2016 04:06 AM, Daniele wrote:
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler.
Do you mean CFQ? http://www.phoronix.com/scan.php?page=news_item&px=cfq-io-scheduler-iops-linux-4.2 Date: 6 June 2015 Title: "http://www.phoronix.com/scan.php?page=news_item&px=cfq-io-scheduler-iops-linux-4.2" Takeaway: "The Linux 4.2 kernel will make the CFQ I/O scheduler default to its IOPS mode when on solid-state drives, which should boost performance." "This change caused the cloud SSD performance in the FIO disk benchmark to increase by more than four times. With a local solid state drive, the performance was about 12% faster under FIO. " -- Life is full of contradictions.... No it isn't. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 04/12/2016 16:09, Anton Aylward ha scritto:
On 12/04/2016 04:06 AM, Daniele wrote:
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler.
Do you mean CFQ? http://www.phoronix.com/scan.php?page=news_item&px=cfq-io-scheduler-iops-linux-4.2
Nope, BFQ ! https://www.phoronix.com/scan.php?page=news_item&px=BFQ-V8-Replacing-CFQ https://software.opensuse.org/package/bfq Bye, Daniele. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Il 04/12/2016 13:15, Daniele ha scritto:
Il 04/12/2016 16:09, Anton Aylward ha scritto:
On 12/04/2016 04:06 AM, Daniele wrote:
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler.
Do you mean CFQ? http://www.phoronix.com/scan.php?page=news_item&px=cfq-io-scheduler-iops-linux-4.2
Nope, BFQ !
https://www.phoronix.com/scan.php?page=news_item&px=BFQ-V8-Replacing-CFQ
https://software.opensuse.org/package/bfq
Bye, Daniele. Daniele, I installed available bfq as kernel module but so far I'm not noticing any impressive performance. I started three consoles, one Virtualbox guest (windowx XP) Thunderbird and Firefox and I observed a long period of HDD high spinning and a bit of system freezing. I still have the parameter vm.swappiness=10 I don't know if I have to set it as per default. Anyhow I will let bfq scheduler activated for a week or two to see if it is better or not for my installation. marco@linux-turion64:~> cat /sys/block/sda/queue/scheduler [bfq] noop deadline cfq
Best regards, Marco
2016-12-04 13:05 GMT-03:00 Marco Calistri <mcalistri@hotmail.com>:
Il 04/12/2016 13:15, Daniele ha scritto:
Il 04/12/2016 16:09, Anton Aylward ha scritto:
On 12/04/2016 04:06 AM, Daniele wrote:
I don't know if is a valid advise for TW but on os 13.2 I am using the BFQ scheduler.
Do you mean CFQ? http://www.phoronix.com/scan.php?page=news_item&px=cfq-io-scheduler-iops-linux-4.2
Nope, BFQ !
https://www.phoronix.com/scan.php?page=news_item&px=BFQ-V8-Replacing-CFQ
https://software.opensuse.org/package/bfq
Bye, Daniele. Daniele, I installed available bfq as kernel module but so far I'm not noticing any impressive performance. I started three consoles, one Virtualbox guest (windowx XP) Thunderbird and Firefox and I observed a long period of HDD high spinning and a bit of system freezing. I still have the parameter vm.swappiness=10 I don't know if I have to set it as per default. Anyhow I will let bfq scheduler activated for a week or two to see if it is better or not for my installation. marco@linux-turion64:~> cat /sys/block/sda/queue/scheduler [bfq] noop deadline cfq
Hi! You forgot to tell us how much memory has your system. Because with less memory system You have higher IO on the hard disk. Regards, Juan -- USA LINUX OPENSUSE QUE ES SOFTWARE LIBRE, NO NECESITAS PIRATEAR NADA Y NI TE VAS A PREOCUPAR MAS POR LOS VIRUS Y SPYWARES: http://www.opensuse.org/es/ Puedes visitar mi blog en: http://jerbes.blogspot.com.ar/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
participants (12)
-
Andrei Dziahel -
Anton Aylward -
Carlos E. R. -
Daniele -
Jan Engelhardt -
Juan Erbes -
Karl Cheng -
Marco Calistri -
Martin Wilck -
Patrick Shanahan -
Robert Munteanu -
Roman Bysh