Hi Robin, On Thu, May 30, 2019 at 10:16:04PM +0100, Robin Hill wrote: > On Thu May 30, 2019 at 08:30:02PM +0000, Andy Smith wrote: > > For testing purposes do you think it would be worth faking it with > > partitions, i.e. 2 partitions on each device making a 4 device > > RAID-10? > > > It depends on what you're trying to test. I don't think it'll work for > performance testing - I think md is smart enough to recognise that the > partitions share the same underlying disk and avoid concurrent > reads/writes. Even if not, you'll just be emphasising the performance > difference of the SSD. Okay, I won't test that then. > Is your interest in RAID-10 for future expansion? I can't see any other > reason not to just use RAID-1 for a 2-device array of solid state > drives. RAID-10's just always been my go-to for a really long time as I'd tested it to be better for my use case back then (more than 10 years ago). So when I tested it again and found it wasn't, I was just interested in why and if that can be resolved. I will go with RAID-1. An expansion is reasonably likely at some point but I don't think I will lose much by just adding it as another RAID-1 pair and then adding that to the LVM VG that the initial array will be used for. Cheers, Andy
