top of page

NASty Thoughts

Backups are important. This is one of the fundamental truths of Computer Science.

Carving things in stone is a reasonably durable medium for information storage. It can wear away or be broken, but if reasonable precautions are taken, it can last for quite a while, though storage capacity is a bit limited... The İmamkullu relief ( is an example of this. It’s a Hittite relief dating from the 13th century BCE, which is a pretty good run.

Unfortunately, computer storage is not quite so durable...

In 1987, I bought my first computer. It was an IBM-XT clone, with a 5 1/4 inch floppy drive, Hercules monochrome graphics, a 1200 baud (external) modem, and a 9-pin dot-matrix printer. And it had a 20MB hard drive.

My storage needs were not excessive, but I never did fill that drive before I moved on to another computer. A bit fuzzier on subsequent machines – you never forget your first – but my next computer, purchased about 3 years later, was a 386SX-25, and I think it had 80MB.

Moore’s law ( states that the number of transistors in a dense integrated circuit doubles about every two years. While it is an observation or prediction, rather than a “law” in any scientific sense, it has held up reasonably well over the past half-century – and not just for transistor density. ( has some other fascinating data on trends in technological progress)

How does this apply to my storage needs over the years? Well, based on the 20MB drive I bought in 1987, and the pair of 4TB drives I just received, I’m not too far off the curve, except that these new drives will be shared for the family, and I’ll end up with 4TB of shared storage, beyond the hard-drives in our various devices.

While most people acknlowledge that backups are important, most people don’t really take it to heart until they’ve been burned a few times. I’ve been reasonably fortunate over the years, but I have often felt guilty about not being sufficiently diligent regarding backups. I’ve backed up files to USB devices, and made multiple copies, but I have never been particularly disciplined about it.

As Graham Cluely pointed out in an episode of Smashing Security devoted to backups (, a process like backing up your computer pretty much has to be easy and automated to work. Humans can be pretty unreliable about things like this, sadly.

In any case, after setting up my Pi-hole recently (, I was browsing lists of Raspberry Pi projects, and came across several references to setting up a Raspberry Pi as a NAS device. Interesting idea, and useful!

Wait! What’s a “NAS device”?

NAS stands for “Network Attached Storage” (, and just means that a storage device is attached to, and accessible from the network. It can range in complexity from a USB key plugged into the back of a router or switch, to dedicated NAS devices with multiple drive bays and a capacity in the dozens or hundreds of terabytes.

In practice, NAS allows for centralized, shared access for users connected to a network, in addition to the hard drives on each device. My plan is to set up a central location for photos and music, automated system backups, and RAID.

Not the bug spray. RAID stands for “Redundant Array of Inexpensive Disks” or “Redundant Array of Independent Disks” (, and simply refers to a number of methods for organizing multiple disk drives in such a way as to improve reliability, availability, performance, and/or capacity.

I may discuss RAID in more detail some other time, but for now, suffice it to say that I’m planning to use “mirroring” or “RAID 1”, which simply means that any data copied to the NAS will be written to two identical drives. So, if either drive fails, I can simply swap it out for a new one, and have the NAS automatically re-build it.

That’s why I’ll be setting up two 4TB drives, and will end up with a total of 4TB of capacity – it’s a tradeoff between reliability and capacity.

I decided I wanted more than just a USB device attached to the router, as it would not give me enough redundancy, and wouldn’t be much fun. I gave serious thought to the Raspberry Pi NAS, as it would be relatively inexpensive and an interesting project, but ultimately decided that I wanted something as robust and easy-to-maintain as possible. (I may still build a Raspberry Pi NAS, but it will be for interest/education, and I probably wouldn’t use it for day to day operations at home – at least not without a LOT more research, and more understanding of how everything works.)

That pretty much leaves me with dedicated NAS devices. Most of my research (, as one example) led me to the conclusion that there are three main decisions to be made:

1) “Populated” or “Diskless”?

NAS devices appear to fall into two main categories – those built by hard-drive manufacturers (who usually use drives they manufacture), and “independent” NAS vendors (who provide lists of supported/recommended drives). After much thought, I decided to go with “Diskless”.

2) Which manufacturer?

This is either a one-part, or two-part question. For a “populated” NAS, the brands I kept seeing were Seagate and WD (Western Digital), both of which have been well-known for their hard-drives for decades. For a “diskless” NAS, I needed to select a manufacturer for both the NAS and the disk drives to populate it.

3) What capacity?

This is normally based on a number of considerations, including how much capacity you need/want, what type of redundancy you want, price-points, and so on. For my needs, the 4TB range seemed to be the best balance between cost, capacity, and reliability.

And the winner is... Well, I don’t know about winner, but I ultimately selected:

The boxes are sitting on the floor behind me, so I’ll be looking into unboxing and setup. I think it’s fair to assume I’ll describe the process, but may take a few weeks to digest everything.




bottom of page