Date: Wed, 1 Jul 2020 15:06:54 +0300 From: Nick Kostirya <nikolay.kostirya@i11.co> To: freebsd-stable@freebsd.org Subject: Re: BerkeleyDB Hash vs Btree on FreeBSD and Linux Message-ID: <20200701150654.74775e56@i11.co> In-Reply-To: <CAKBkRUwnc9aC00NasdMX=g8n1jhiXvfPDwjKaz2sYbB4Hspk5g@mail.gmail.com> References: <20200701130649.0b0643d5@i11.co> <CAKBkRUwnc9aC00NasdMX=g8n1jhiXvfPDwjKaz2sYbB4Hspk5g@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 1 Jul 2020 18:20:19 +0800
Li-Wen Hsu <lwhsu@freebsd.org> wrote:
> On Wed, Jul 1, 2020 at 6:07 PM Nick Kostirya via freebsd-stable
> <freebsd-stable@freebsd.org> wrote:
> >
> > Hello.
> >
> > I noticed that BerkeleyDB Hash is VERY slow compared to BerkeleyDB Btree on FreeBSD (UFS or ZFS).
> > But they (Hash and Btree) have roughly the same performance on Linux.
> >
> > Why?
>
> It's not an easy question, do you have more information about the test
> environment setup, and the statistics of the results?
>
> I'd recommend using some analysis tools like DTrace to check what it's busy for.
The top show getblk status often.
Please tell me what you can and how to look with DTrace.
I use
dtrace -n '::: /execname == "a.out"/ { @[probefunc] = count(); }'
but I do not see the difference between Hash and Btree.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20200701150654.74775e56>