how a shingled WD Red is benchmarking
The dips in read speed seem to indicate the disk-managed shingle management kicking in
Although 128MB/sec isn't stellar for write speed, it's way faster than the 40MB/sec I'm allegedly supposed to be topping out at with these drives.
(it's a single disk in a USB 3.0 external SATA hookup)
@yakkoj Aren't they supposed to get really bad when they are nearly full? Maybe they aren't fully shingled.
@gudenau they apparently get really bad when the non-shingled buffer zone needs to be flushed out to shingled zones. Word on the street is that they'll be fine resilvering for about 40 minutes, and then block for so long that zfs times it out and kicks it from the array
As resilvering at 4TB is about 3 hours without additional activity here, I couldn't use these on my arrays at all.
@yakkoj Ah, that makes sense.
I wonder if there's like 20 shingled secters then a break so it doesn't have to do the entire platter.
This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!