That shouldn't be a problem. The robustness of keeping the user/docs data sets separate from the zones is in convenience of upgrading the OS in the zones without affecting the data. If something goes wrong - do zfs rollback on the zone. I have 3 pools on my production system, Fujitsu RX300s7, filled with 6 x 7200 RPM 1TB SATA for almost a year: vol0 - mirrored for Maildir. vol1 - raidz1 for www (htdocs, mysql db - for multiple apache CMS/Drupal web sites). zones - for 3 smartos zones (one with postfix e-mail server, and two with apache/php/mysql web servers). Each zone has its on GigE assigned. Both vol0 and vol1, I/O intensive, are given to the zones through lofs. The performance was very good.
Regards, Alexei > Greetings all, > > I am planning to migrate my primary local file and zones server to > SmartOS, to match my datacenter machine configuration. (And because I > generally just like SmartOS.) I currently have a two pool setup (ignoring > rpool): A large, relatively slow pool consisting of enterprise-rated 3.5 > 7200 RPM SATA disks, and a fast, smaller pool consisting of 2.5 15k SAS > disks. I put backups and large media files on the large pool, and home > directories, databases, working set files, and such on the fast pool. > > Im trying to reduce power consumption, so Im about to upgrade the > machine to a Haswell processor and lower-power disk chassis. I may try to > reduce the number of disks, as well. I did some work some time ago to > reduce a number of physical machines to just one with a bunch of zones, > and Id like to keep it down to a single machine, both for power and cost. > (I keep offline parts for failure recovery, and SmartOS makes this really > easy.) > > My question is: What is the current wisdom on translating this to SmartOS? > I seem to recall some advice long ago on this list against having more > than one pool on a SmartOS machine. Other than splitting the setup into > two machines, one with a fast pool and one with a large pool, is there any > general advice? Do people find a properly set up pool of 7200 RPM disks to > be reasonably performant these days? I know even a small pool of such > disks can easily saturate a GigE link; Im more concerned with latency > traversing hundreds or thousands of directories and returning metadata and > small bits of data to clients (via NFS, web services, etc.) Do I just need > gobs of RAM to solve that problem? > > Does anyone combine such uses into a single machine? How much pain would I > be in for if I tried to go for a two-pool setup? > > - Geoff > > > > ------------------------------------------- > smartos-discuss > Archives: https://www.listbox.com/member/archive/184463/=now > RSS Feed: > https://www.listbox.com/member/archive/rss/184463/26842518-03f78edb > Modify Your Subscription: > https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com > ------------------------------------------- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
