Date: Mon, 15 Nov 2010 17:11:52 +0100 (CET) From: Oliver Fromme <olli@lurza.secnetix.de> To: freebsd-geom@FreeBSD.ORG, bsemene@cyanide-studio.com Subject: Re: Newbie RAID 0 question Message-ID: <201011151611.oAFGBqUo011222@lurza.secnetix.de> In-Reply-To: <4CE0FD0B.2010700@cyanide-studio.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Bastien Semene <bsemene@cyanide-studio.com> wrote: > If I'm understanding well, I can add space with gconcat (8) but there > will be two separate stripings, in your example datas will be stripped > between ad0/ad1 and when space is full between ad2/ad3 ? Yes. > I'm using RAID-1+0 structure, so I think I can do the work with 2 arrays > (old one with datas, new one with datas and more space). You didn't mention RAID-1+0 in your first posting, only RAID-0. That's quite a difference. If you expect to receive good advice, you should describe your existing setup as precisely as possible. > Do you see some advantage/disadvantage using the gconcat model instead > of a true striping ? Both is "true" striping, there's only a difference in how the data is organized, which can have an impact on performance. When four disks are organized in a single RAID-0, this means that stripes are distributed equally across all four disks. In theory this can improve performance, especially for sequential access. When two pairs of disks are organized in two RAID-0 volumes which are then concatenated, the stripes for each of the sub-volumes are distributed only across two disks. Therefore there's a lower limit on the maximum performance that can be achieved (in theory). In other words: Adding two new disks using a concat container will not make the whole thing faster (but not slower either). It will be just bigger. It should also be noted that the new volume will be less reliable. Every time you add another disk to an array without redundancy (no matter whether RAID-0 or concat), you reduce reliability, because the whole array will go down when any one of the disks fails. The more disks, the higher the probability that one of the disks fails. Of course, if you have RAID-1+0, then things are a lot different. > Is it planned that geom will add an insert command, or is gconcat the > definitive solution ? I'm not a geom developer, so I'm not authoritative, but I don't think that such a feature is planned. As I wrote previously, adding another disk to an existing RAID-0 is technically expensive and time- consuming. I don't think it's worth the effort. Best regards Oliver -- Oliver Fromme, secnetix GmbH & Co. KG, Marktplatz 29, 85567 Grafing b. M. Handelsregister: Registergericht Muenchen, HRA 74606, Geschäftsfuehrung: secnetix Verwaltungsgesellsch. mbH, Handelsregister: Registergericht Mün- chen, HRB 125758, Geschäftsführer: Maik Bachmann, Olaf Erb, Ralf Gebhart FreeBSD-Dienstleistungen, -Produkte und mehr: http://www.secnetix.de/bsd "The ITU has offered the IETF formal alignment with its corresponding technology, Penguins, but that won't fly." -- RFC 2549
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201011151611.oAFGBqUo011222>