Trying to understand bonnie++ output / 2023-08-07

2023-08-07 Trying to understand bonnie++ output
In preparation for a migration at work I wanted to do actual benchmarking of Linux filesystem performance. I think I used bonnie in the last century, so I gave bonnie++ a spin for this.

I have little idea of what 'good' or 'bad' numbers are from bonnie++. I could only compare a "local" filesystem with an NFS filesystem. I use local in quotes because this was in a virtual machine, so it's SSD storage in raid-1, with LVM on top of it, with a logical volume assigned to a KVM-based virtual machine, which uses the virtio disk driver for an ext4 filesystem.

The numbers for the "local" filesystem:
Version  1.98       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Name:Size etc        /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
gosper          32G  809k  98  440m  38  215m  22 1590k  99  410m  30  4639 135
Latency             25688us     317ms     143ms    9332us   39208us    2089us
Version  1.98       ------Sequential Create------ --------Random Create--------
gosper              -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
Latency               488us     684us     762us     236us      87us     262us
1.98,1.98,gosper,1,1691401899,32G,,8192,5,809,98,450230,38,220385,22,1590,99,419827,30,4639,135,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,25688us,317ms,143ms,9332us,39208us,2089us,488us,684us,762us,236us,87us,262us
And for NFS, a Synology NAS with spinning disks in raid-5:
Version  1.98       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Name:Size etc        /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
gosper          32G 1054k  98 78.7m   7 68.4m  13 1483k  99  109m  10 432.2  12
Latency             11138us     408ms   13261ms   16434us     212ms     274ms
Version  1.98       ------Sequential Create------ --------Random Create--------
gosper              -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 16384  10 16384  16 16384  15 16384   9 16384  18 16384  15
Latency             69194us   53194us   98927us   69144us    1240us   94317us
1.98,1.98,gosper,1,1691398200,32G,,8192,5,1054,98,80605,7,70058,13,1483,99,111574,10,432.2,12,16,,,,,705,10,16184,16,2800,15,693,9,3703,18,2226,15,11138us,408ms,13261ms,16434us,212ms,274ms,69194us,53194us,98927us,69144us,1240us,94317us
Now I am somewhat confused. Sequential write to NFS is slightly faster.

Update 2023-08-08

At work I got different but comparable numbers for iscsi attached storage versus vmware storage (and the layers in between). Those numbers helped make decisions about the storage.

Tags: , , ,

IPv6 check

Running test...
, reachable as koos+website@idefix.net. PGP encrypted e-mail preferred. PGP key 5BA9 368B E6F3 34E4 local copy PGP key 5BA9 368B E6F3 34E4 via keyservers

RSS
Meningen zijn die van mezelf, wat ik schrijf is beschermd door auteursrecht. Sommige publicaties bevatten een expliciete vermelding dat ze ongevraagd gedeeld mogen worden.
My opinions are my own, what I write is protected by copyrights. Some publications contain an explicit license statement which allows sharing without asking permission.
Other webprojects: Camp Wireless, wireless Internet access at campsites
This page generated by $Id: newsitem.cgi,v 1.62 2023/09/19 14:49:50 koos Exp $ in 0.009997 seconds.