Date: Fri, 5 Feb 2021 15:45:42 -0500 From: Gunther Schadow <raj@gusw.net> To: freebsd-performance@freebsd.org Subject: Re: FreeBSD on Amazon AWS EC2 long standing performance problems Message-ID: <3fde2934-1e18-5ea4-84d6-21200eaf4b20@gusw.net> In-Reply-To: <YB2LoIWBZH8P%2BQn2@lion.0xfce3.net> References: <mailman.0.1612528010.40441.freebsd-performance@freebsd.org> <98fc52d4-caf1-8d48-5cb2-94643a490d4f@gusw.net> <YB2LoIWBZH8P%2BQn2@lion.0xfce3.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Gordon Bergling wrote: > Can you verify your feelings by numbers? Yes, like I said >> Not by a few % points, but by factors if not an order of magnitude! https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=253261 Do this: dd if=/dev/zero of=/dev/nvd2 bs=100M status=progress and you see that it's writing with the "whopping" speed of 70 MB/s. That used to be good, but it is no longer good. Compare Amazon Linux doing the same thing at 300 MB/s. Now, when you put a file system over it, zfs or ufs, then instantly the performance gets better: newfs /dev/nvd2 mount /dev/nvd2 /mnt dd if=/dev/zero of=/mnt/test bs=100M status=progress now that works at about 250 MB/s. Decent. So, problem solved? No! It turns out if I create a PostgreSQL database over this setup, then again there is massive delay on the read and write and throughput will drop to even worse than 70 MB/s. Creating one index takes 10 times as long as that same on the Linux system. PS: no need to point out that Linux uses buffer cache for direct write to device and BSD doesn't. Those effects will not make a difference when you write (or read) more than the buffer cache size (e.g., a few GBs).
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3fde2934-1e18-5ea4-84d6-21200eaf4b20>