Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 4 Apr 2016 21:38:16 -0400
From:      "Tenzin W. Lhakhang" <tenzin.lhakhang@gmail.com>
To:        Miroslav Lachman <000.fbsd@quip.cz>
Cc:        Wim Lewis <wiml@omnigroup.com>, "freebsd-fs@FreeBSD.org" <freebsd-fs@freebsd.org>
Subject:   Re: ZFS pool with a large number of filesystems
Message-ID:  <D7F2C273-66A6-49B5-B653-16122910DEB4@gmail.com>
In-Reply-To: <570311C5.4010702@quip.cz>
References:  <34DB45E8-7E1F-4D7C-96FF-E0A403EE8000@omnigroup.com> <570311C5.4010702@quip.cz>

next in thread | previous in thread | raw e-mail | index | archive | help
The most I have seen is approximately 10,000.

The performance isn't too bad at that scale.  The clone fs metadata becomes a=
 bit costly.  I seen pretty deep nesting into the zfs dataset chains on cert=
ain server types; dataset clone snap clone snap chains going about 30 levels=
.

Note:  we are running zfs on illumos (Smartos)

General system specs are:  e5-26[89]0v2,256gb ram, 1-2 ZIL and ssd or spinni=
ng with 4-6 ssd cache drives.

Tenzin

Sent from my iPhone

> On Apr 4, 2016, at 9:15 PM, Miroslav Lachman <000.fbsd@quip.cz> wrote:
>=20
> Wim Lewis wrote on 04/05/2016 02:38:
>> I'm curious how many ZFS filesystems are reasonable to have on a single m=
achine (in a single zpool). We're contemplating a design in which we'd have t=
ens of thousands, perhaps a couple hundred thousand, filesystems mounted out=
 of the same pool. Before we go too far into investigating this idea: Does a=
nyone have real-world experience doing something like that? Is it a situatio=
n that ZFS-on-FreeBSD is engineered to handle with good performance? Is ther=
e a rough estimate of the resources consumed per additional filesystem (in t=
erms of kernel VM and disk space)?
>>=20
>> Thanks for any insight or advice (even, or especially, if the answer is "=
that's crazy, don't do that" :) )
>=20
> I donn't know about how many filesystems but I know that few hundereds of s=
napshots can make a noticeable slowdown for some zfs operations.
> I think that basic "zfs list" will be painfully slow with tens of thousand=
s of filesystems.
>=20
> Miroslav Lachman
> _______________________________________________
> freebsd-fs@freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-fs
> To unsubscribe, send any mail to "freebsd-fs-unsubscribe@freebsd.org"



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?D7F2C273-66A6-49B5-B653-16122910DEB4>