I bought a ReadyNAS NV+ years ago, and don't like it. It serves files, but does so many things badly that next time I want to add more disk space, I intend to purchase something else.
The (three) disks filled up. I wanted to replace them. The natural procedure: Insert a larger disk into the spare slot, copy the contents of one of the old disks onto the new disk, switch over so the RAID uses the new disk instead of the old, remove the old. Repeat twice more. After that, resize the fs since the disks are larger.
No such luck. The ReadyNAS way is much riskier: Remove a working disk and let the RAID subsystem regenerate its contents from the other disks. A power failure would be undesirable in the ten hours after swapping a disk, so said the readynas. Three unnecessary ten-hour windows of bad luck: Not impressive.
Does it really need to take five seconds to load simple web page across a gigabit ethernet?
In general the web interface is clunky beyond belief. I added as much RAM as the box would take, hoping that would speed things up. No. It really does take five seconds to serve a login prompt.
No support from the ReadyNAS, and as far as I can tell none will be forthcoming in the foreseeable future.
I have three streaming clients (two FLAC/MP3 players and one video player). For my next NAS, I want streaming to work without jitter even if other clients are using the NAS.
(Update: When I disable the builtin Squeezebox server, performance improves drastically. Videos play without jitter, and the web server is almost acceptably fast. It seems the built-in mysql server is to blame. It eats the CPU even when nothing uses the database. My song database is/was modest, with about 20,000 rows in the biggest table.)
Performance in general
Making backups to an USB disk is much too slow (I get less than 100G per day). NFS support is also not exactly stellar.
The ReadyNAS will spin the disks down based on idleness in the past minutes: The most primitive algorithm known to mankind. For my next NAS, I at least want the ability to plug in my own algorithm (
record activity in five-minute chunks for two weeks; spin down if the coming hours were idle in the past days, perhaps modified to distinguish the working days from the weekend), better would be clever spindown without having to implement it myself.