NAS Server Final Thoughts

My first and solemn duty is to remind everyone that relying on a collection of drives in any RAID configuration for data backup purposes is a huge error. RAID systems provide protection against loss of services, not loss of data. Several examples will illustrate the problem, I hope:

• the drive controller goes bad and corrupts the data on all the drives in the array
• the entire storage device is physically or electrically damaged by external forces
• the entire storage device is lost, stolen, or destroyed
• a single drive in a RAID 5 cluster dies and during the rebuild process, which puts higher stress on the remaining drives, a second drive fails
• floods, earthquakes, tornadoes, etc. (AKA El Niño)

All these points lead to the inescapable conclusion that multiple drives in a common system, in a single location do not provide effective and reliable data backup. Throughout many of the NAS reviews I've talked about high-availability systems, but the NetDISK 351UNE doesn't quite fit that description. However, if you bundle two or three units together, and use the binding features available in the software, you can create a RAID 1 (+ hot spare) volume across the seemingly individual units. That's a unique capability that gets pretty close to enterprise-level redundancy, but with only one unit on hand, I couldn't test it. What you can't get, are the performance advantages of the higher-level RAID configurations. What you do get in exchange is additional system-level redundancy, because each NetDISK box has its own power supply and its own electronics. In a typical NAS, if one electronic component in the PSU or on the main system board fails, the whole unit goes down.

So, even if one whole NetDISK unit goes up in a puff of smoke, your data is still available and accessible. The virtual RAID volume stays online the entire time while the failed device is replaced and the array is rebuilt. That's what RAID systems are designed to do. The inherent redundancy is not meant to serve as a backup file set. A larger NAS with more drive bays offers the possibility of increasing the redundancy with RAID 5/6/50/60, but for a SOHO environment, RAID 1 or 5 is generally adequate. Remember, we're not talking about losing data here; we're only talking about the ability to keep working uninterrupted, if one drive should fail. If two or more drives fail, it's time to pull the local backups off the shelf.

The question I keep coming back to is, "Is there a specific role that a small NDAS device fulfills so well, that it displaces all the other options?" I realize I'm spoiled for life after my last round of NAS testing, but I can't see relying on a single drive for front-line data storage, after living with RAID 5 for a while now. I've been through the process of restoring a system from backups, and if I can avoid it, I will. Even just restoring user data is a challenge; because you have to know exactly where to put it back to, and then the applications have to know that it's there. That's why drive images and snapshots are so popular in the enterprise world. You make an exact clone of your system drive, and put it on the shelf. Of course the replacement process works best when you have an identical workstation sitting there on the same shelf, because the drive image has all the device drivers and system settings that are unique to the hardware it was generated with.

How about as a backup device? That's where I think the NetDISK 351UNE really shines. With all the major interfaces available in one unit, it makes for great flexibility during both the backup and restore process. It's small enough to be transportable, and it supports the new 3GB SATA drives, so it will more than likely hold all the precious data you need to keep on ice. With a simpler (direct-attached-storage) device, you would typically swap two or maybe three units around, always keeping one at a remote location. With an NDAS, you can set up a virtual RAID 1 volume, and keep the array mounted to the network while you swap out the hot spare. That way, you can keep the same drive letter assigned in your desktop environments and not have to configure the drive anew each time it's brought on-line. The replacement drive starts synchronizing itself to the primary disk as soon as it's attached. There are cheaper solutions for straight backup applications, with eSATA and USB 3.0 connections, but it's a whole lot harder to make them accessible to every workstation on the network.

So, what conclusions can we draw, particularly about this high value, single-bay NDAS device? Click NEXT to find out, and discuss...

<< Previous Page - Next Page >>