Date: Fri, 22 Jun 2012 08:22:42 +0200 (CEST) From: Wojciech Puchar <wojtek@wojtek.tensor.gdynia.pl> To: kpneal@pobox.com Cc: Matthias Gamsjager <mgamsjager@gmail.com>, FreeBSD Questions <freebsd-questions@freebsd.org> Subject: Re: Is ZFS production ready? Message-ID: <alpine.BSF.2.00.1206220817350.26433@wojtek.tensor.gdynia.pl> In-Reply-To: <20120622024012.GF25628@neutralgood.org> References: <CAPj0R5Kmi-%2BdJ7mPvTrTAoS8O983svOyR2WyK2_v1Cr07dSS_A@mail.gmail.com> <alpine.BSF.2.00.1206211413140.2263@wojtek.tensor.gdynia.pl> <CA%2BD9QhuQ%2BbxKW9%2BdX%2BzS9mErwz8JSkV2G7qL0KfB8BH_LGJAgA@mail.gmail.com> <alpine.BSF.2.00.1206211539230.2903@wojtek.tensor.gdynia.pl> <CA%2BD9QhvR_eKtVxdKcaMyOS7tLw_AOHKgUy3o7mJn2b=chMA0Xw@mail.gmail.com> <alpine.BSF.2.00.1206211619250.3092@wojtek.tensor.gdynia.pl> <CA%2BD9QhvwKZm7heoe7tpfhYCJvkknw_HC7aFjCu%2B-1xYQBmV6ng@mail.gmail.com> <alpine.BSF.2.00.1206211644350.3170@wojtek.tensor.gdynia.pl> <CA%2BD9QhsyOh34SghWzQPpnTig%2BUmSEO2VP7jfPxTXs9zW9Uakeg@mail.gmail.com> <alpine.BSF.2.00.1206211707570.3361@wojtek.tensor.gdynia.pl> <20120622024012.GF25628@neutralgood.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> OK, if you have 24 2-way mirrors and two drives in the same mirror fail > then with UFS you lose the contents of that mirror. Other filesystems in > the same box are fine. Restores from backups are going to be easy since > the backups are probably arranged to be per-filesystem. true. i actually don't have 48-disk machine but do have 9 disks (one SSD+8 2TB SATA). > So far I think we're in agreement. Still as i said - even with ZFS i would make 24 pools, not one. this thing is not filesystem dependent. > > But this doesn't address two issues: > > 1) There are other arrangements of ZFS that can tolerate more failed > disks if you are willing to spend more money. ZFS supports n-way > mirrors, so you can have mirrors with three or four disks if you as well as gmirror. > a raidz2 set (with multiple raidz2 sets per pool). i will not use raidz1/2/3 because if catastrophically low performance. the design of ZFS makes sure you'll get read performance of single drive from whole pool. Disks are already performance limiting part of computer. > 2) That this failure can happen doesn't address the question of the > production-ready status of ZFS. > > The question of "production ready" is not a boolean. It is a question of What i meant from beginning is not that ZFS is not "yet" production ready but it will never be because of design decisions. It have "cool" features, giving danger, huge hardware usage (RAM,CPU) and low I/O performance. > risks and of money used to mitigate those risks. I suggest asking the > question on the zfs-discuss list over at opensolaris.org since there are > probably many more people there who make serious use of ZFS daily. I will not. Serious people should know how ZFS work. if they still want to use it seriuosly then i cannot help any more. > gs1p 159G 73.1G 39 12 2.34M 70.7K > mirror 159G 73.1G 39 12 2.34M 70.7K > gpt/CONST_2-9XE02KPK-zfs - - 19 5 1.94M 69.4K > gpt/SAVVIO-6XQ10F80-zfs - - 21 5 1.93M 79.5K > gpt/SAVVIO-6XQ103C7-zfs - - 21 5 1.93M 79.5K > 100GB+ of FreeBSD being served up (IP 206.196.19.100 if you care to check > FreeBSD's stats pages). And the torrents can be easily replaced if something > really bad happens. 3 very expensive drives.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.1206220817350.26433>