As I said above, I bought a 4-bay 1100. After building the 3-drive RAID5 array with a hot-spare, I removed one of the drives. It immediately started to rebuild the array with the hot-spare. 

So I removed the hot-spare and inserted it again. As the author observed, it wouldn't rebuild the array anymore, even after inserting both removed drives.

Perusing the Infrant forum, I noticed that the firmware doesn't expect immediate insertion of drives back to back or quick removal-insertion cycles.

So after rebuilding the array again, I pulled a drive and as expected, it started to rebuild the array with the hot-spare. While it was still doing it, I removed the hot-spare and it went into non-redundant mode, as expected. I reinserted one drive and it still wouldn't rebuild the array, advising a reboot. 8-O

So I did and the array started to be rebuilt again. I inserted the hot-spare back, but it wasn't automatically added as a hot-spare, requiring me to do that manually.

So, it's sort of redundant. But it should work in typical failure conditions, when only one drive fails at a time.

I wish that the hot-spare would be spared usage, but it just so happens that the OS is mirrored through all drives installed, which means that the hot-spare is only a hot-spare for the user's volume. No wonder several users reported difficulty getting the drives to spin down at idle.

I'm sorry but the author of the write up did not state that they installed a spare drive during the raid 5 test. He only stated he removed the drive and it failed to rebuild. I own two of these systems and have tested the RAID 5 rebuild repeatedly. IT HAS NEVER FAILED.

PERIOD.

So let me recap... a disk is removed and re-added to simulate a disk failure and it fails to rebuild the array ? Is this a joke ? Why do so many people BLINDLY trust this storage appliance when the most basic reliability feature supposedly provided by a raid array doesn't even WORK ?

For Unix users, I would rather recommend Linux MD/LVM or better: ZFS, which supports even more features than the ReadyNAS: end-to-end checksumming, self-healing, scrubbing, snapshots, clones, compression, pooled storage, pool growing, simple admin, instantaneous array init, always consistent on disk, etc.

For inexperienced users, I am afraid there is no easy solution for them, other than putting up with "products" like the ReadyNAS... :(

Is this article bashing this box because it sucks in RAID 5? This is absolutely bizarre. This machine is built for RAIDX. I absolutely love this box. All it is missing is eSATA for backups and some better performance when viewing folders from Windows.

I have had my box for two years. Bravo NETGEAR or whoever did the latest firmware. A much better interface, bigger hard drives, my WD USB MyBook no longer hangs the system. Truly an awesome firmware for an awesome machine.


I've been using one for a couple of years (2x500) and although I've not been disappointed in the performance or stability, the firmware could certainly use some improvements; namely, an integrated authentication server (as stated above) and a syslog collector.

Even just the auth portion would allow it to master a domain, which would satisfy the needs of a great number of SMB and home users.

I just purchased its 1U cousin, the ReadyNAS 1100.

Even though X-RAID is attractive, I really intend to use RAID-5 on this unit because of the extra reliability of having a hot-spare, albeit at the cost of less space.

But I'll try to perform reliability tests to verify that it can rebuild the array properly. Otherwise, hot-spare or not, it's not reliable at all.

Good review and yes this device is quite quirky. We are a ~150 desktop local gov shop and have an NV+ from before the Netgear acquisition. We use the X-RAID format due to issues you mentioned ;) Another funny low point is their recommendation to reverse the flow of the cooling fan to improve cooling of the system board. What they failed to consider is that now the intake air is not filtered and dust buildup will be a problem. We decided to take our chances with thermal failure and not do the fan reversal, esp. since it operates in a A/C controlled server room.

We use this device as a cheap disk->disk NL storage and backup. One huge plus you forgot to mention is the rsync capability (hopefully not gone in the latest release). This can save a good deal of bandwidth when backing up things over a slower WAN connection.

Also, a close scouring of the forums will reveal that the CPU in this device is a very slow and old 32-bit SPARC. It's too bad, since a modern ARM CPU would likely be much faster. People have upgraded the RAM to 1GB to improve performance.

It's been awhile since I futzed with Linux, but from trawling the logs I believe their RAID is just software LVM RAID. I doubt that sparc has the muscle to do HW RAID5 off a 1Gb connection anyway so no real loss. Their "proprietary X-RAID" may not even be proprietary at all...might just be some interesting scripts that control rebuilding and servicing plain jane raid levels. It's very interesting to see the benchmarked differences in performance.

Now that they've officially added the SSH access I hope that some alternative firmware can be created, much like the OpenWRT series for routers.

One of my companies has used an NV+ (same configuration: 4x250MB) for the past nine months. In that time, it has behaved faultlessly. We use X-Raid, and don't see the point in anything else: after all, that's what Infrant/NetGear designed it for, and it's not as if you'd use the disk set in any other device.

The major win as far as we are concerned is NFS (Network File System) support, which just about no other RAIDs in its class has. This is essential for an office that has a mix of Linux/Unix, Mac and Windows machines. Sorry but CIFS alone just doesn't cut it in a professional mixed-OS environment.

Our only complaint is that, as supplied, the NV+ doesn't operate as a heterogeneous authentication (RADIUS) server as well, which would make it a one-box machine room for many companies.