New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Never heard of it, but now that I have, I will
Thank you!
Hi Mark. I opted for just caching the reads as most of the workloads my current customers have are read oriented. Now, bear in mind this is just v1.0 of vRocket and we had to start somewhere.
I am looking into introducing ZIL devices into all of our nodes within a couple of months (great thing about ZFS as you can add such things whenever you choose), but for now, the only scenario I've personally seen them as absolutely necessary, is running Windows in KVM.
As for RaidZ1, reason is simple, I have 2 hot spares per host, and over 50 of the spare drives at the location which can be installed by me or remote-hands in just 5 minutes if there's need.
Thanks.
I've been playing with all 3 variants of their RAID and RAIDZ1 was slow at a rebuild, RAIDZ2 with a single disk failure was faster to rebuild. During that rebuild process, IO performance was in the toilet, I just wondered if there was a specific reason that you'd discovered during testing.
Yea, resilvering can take a moment. In the long run we'll be 100% SSD and w/ ZFS we can do that upgrade live on the host - one drive at a time
Make sure that you are running logbias=throughtput if you are running an all-SSD zpool.