Date: Tue, 12 Jul 2022 09:22:48 +0100 From: Frank Leonhardt <freebsd-doc@fjl.co.uk> To: questions@freebsd.org Subject: Re: Effect of using SATA 2.x and 3.x in zfs mirrored config Message-ID: <dce70693-6556-ae09-d21d-85e5e5cff7f7@fjl.co.uk> In-Reply-To: <1406347038.59069302.1654294344615.JavaMail.zimbra@shaw.ca> References: <1706301853.51806224.1654176891423.JavaMail.zimbra@shaw.ca> <20220602220732.358b6b81b07d35ac608c88b1@sohara.org> <1406347038.59069302.1654294344615.JavaMail.zimbra@shaw.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
On 03/06/2022 23:12, Dale Scott wrote: > ----- Original Message ----- >> From: "Steve O'Hara-Smith" <steve@sohara.org> >> To: "Dale Scott (dalescott@shaw)" <dalescott@shaw.ca> >> Cc: "freebsd-questions" <freebsd-questions@freebsd.org> >> Sent: Thursday, June 2, 2022 3:07:32 PM >> Subject: Re: Effect of using SATA 2.x and 3.x in zfs mirrored config >> On Thu, 2 Jun 2022 07:34:51 -0600 (MDT) >> Dale Scott <dalescott@shaw.ca> wrote: >> >>> Looking in dmesg, I see one of the drives is using SATA 3 (600MB/s >>> transfer), and the other is using SATA 2 (300MB/s transfer). Will mixing >>> SATA 2 and SATA 3 drives cause any technical issues? Is there any reason >>> why I should use SATA 2 for both drives? >> No reason at all to do so, but it probably wouldn't hurt either, I >> doubt the drives can exceed SATA 2 speeds for long if at all. >> >> ZFS won't know or care what underpins the devices. I've even >> mirrored a laptop SSD with an iSCSI accessed zvol on a NAS - it worked fine >> (I didn't try disaster recovery, that would take some thinking about) the >> main aim was to make a weekly scrub worthwhile. >> >> -- >> Steve O'Hara-Smith <steve@sohara.org> > > Thanks for vote of confidence in ZFS. Since JBOD works I wasn't expecting trouble, but it was still a surprise to learn the drives were using different protocols. My larger concern was, if there was a reason why using the same protocol was "better", whether moving the SATA cable to a different port would trigger issues with ZFS, and I would need to learn much more about ZFS than I do now. > > Cheers, > Dale > Steve et al are correct. ZFS is very robust and mismatched drive performance is not an issue. With mechanical drives you expect a mismatch (access time) and they get worse over time, and different drives are always being added to arrays via all types of HBAs. Newer drives tend to be faster There are questions, however. For example, does a synchronous write drop to the speed of the slowest drive in a mirror? POSIX demands the data is in non-volatile storage, but one copy should be enough and that could be in the zil rather than it's final resting place. It's obviously going to affect performance but I doubt you'll notice. Regards, Frank.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?dce70693-6556-ae09-d21d-85e5e5cff7f7>