Date: Mon, 17 Mar 2014 15:42:10 +0100 From: n j <nino80@gmail.com> To: User Questions <freebsd-questions@freebsd.org> Subject: Monitor IO performance Message-ID: <CALf6cgY-if5RH9y19aQf8DTV=6Kkux%2B18H5aHkMNsJK-knEMUw@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hi, I'm having issues with IO performance on a 10.0-RELEASE VPS. I suspect my provider might have oversold the resources. I'd like to collect some statistics before I send them a complaint. I'd welcome any help or pointers to links/documentation on how to best test IO performance and collect IO statistics on a prolonged period (e.g. run something every 2 hours for a few days or something like that) and actually make some sense out of those numbers. Right now, I'm thinking about running ioping (/usr/ports/sysutils/ioping) as I find its output quite clear, but I'm not sure how good an indicator it is without a benchmark to compare it to. Here's a sample run, these numbers look abysmally low to me: # ioping -R /tmp --- /tmp (ufs /dev/ada0s1a) ioping statistics --- 66 requests completed in 3.0 s, 21 iops, 87.8 KiB/s min/avg/max/mdev = 756 us / 45.6 ms / 681.1 ms / 128.2 ms P.S. The problems started some time ago and might also be related to the 10.0-RELEASE upgrade. I can't really pinpoint if the problems started with the upgrade or not, but would appreciate info if anyone else experienced any IO performance degradation after the upgrade. Thanks, -- Nino
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALf6cgY-if5RH9y19aQf8DTV=6Kkux%2B18H5aHkMNsJK-knEMUw>