Why tapes are still the most efficient backup method?

For businesses it is obvious that the vast amount of data (at the time of writing this, tens and hundreds of gigabytes, if not terabytes) make high capacity tapes the medium of choice for backup.

Large data centers cannot live without tape libraries AND offsite tape storage and rotation, to cover against the case of a disaster.

Believe it or not, tape drives are also the best backup for a home system as well.

Of course, some people will jump in right now and object to tape drives with various arguments. These arguments are mostly flawed as per the discussion below. Most of them come from people who only have experience around PCs or PC operating systems (e.g. Windows), and no exposure to how mainframes/mini-computers used to be, nor how data centers are run.

Let us address them one by one:

Objection Comments
Who needs backup when you have RAID RAID is local to one machine. If that machine catches fire, or gets stolen, the data is no where to be found. Moreover, RAID does not protect you from user error. If you delete file or a directory by mistake, RAID will not help you one bit. Only a backup on a separate media / directory will help with these "Oops!" mishaps.
Tape is old and obsolete technology! Tapes are definitely old as a concept. But the technology keeps evolving, with more capacity and speed all the time. There are also robotic tape libraries that handle the insertion / removal of tapes from drives, although these are beyond home use.
Optical media such as CD-Rs and DVDs are better! Optical media now are either CD-Rs (approx 700 MB each?) or DVDs (4 GB each?). These capacities are not enough to backup an entire disk today. Moreover, the media is mostly write once, so it is throwaway, unless you use CD-RW, which are too small, or DVD RW, which is still nascent technology with high cost drives and media, as well as low capacity compared to tape.
IDE disks are so cheap. Just buy another disk, and put in the computer and use it only for backup. While that helps a bit, it is still flawed. If the original hard disk fails, this will protect you. However, if the power supply blows up and fries all components and drives, you are out of luck. You are also out of luck if you get your basement flooded or have a fire.
Firewire external drives or IDE hard drives in a caddy are better and have more capacity These are certainly a step above the RAID argument or the optical media one. They will help you with the scenario where the server power supply catches fire. They will not help with other scenarios though. If you keep the disk at home, then a fire at home destroys your data AND the backup of that data. Moreover, you only have one copy of the data, so if you delete a file by mistake, then do a backup, that backup does not have the file on it.
rsync mirroring does the job for me It depends on how you set it up. If you set it up to a remote machine (offsite, at work, university or at a friends place), then it is much better than the above options. However, do you keep multiple versions of the backup for a month or more? If so, then congratulations, and you are lucky to have all that space and resources! If not, then you have the same drawbacks as above. A local rsync over the LAN is relatively fast, but again, it is in the same home, subject to theft, flooding or fire.

You can read more on how I use tapes on Linux to backup a home network, where are there are lots of details on the hardware, configuration, and commands used, as well as lots of links to various information on backup.



What about the obvious

What about the obvious solution of *two* external Firewire drives, with one always in a different location?


That is a viable solution.

The only drawback is that cost per megabyte is more than that of tape still. If you need incremental daily backups and then weekly full backups, with one offsite, it adds up.

However, with disk drive capacity going the way they are, there are no longer affordable tape drives for home/small business use that keep up with them.

I am about to outgrow my tape backup solution, since it has 10 GB capacity, and I already do two backups (one for photos, and one for everything else).

It seems that a USB 2.0 drive enclosure is the most convenient way. I saw some for $16 or so. The ones for 2.5" drives does not require external power even, but the drives are more expensive. My brother uses one of those.

Here is someone who built his own USB 2.0 drive enclosure with cooling fans.
Khalid Baheyeldin

But you still have the cost

But you still have the cost of the original tape drive for large capacities, which is very expensive. You have to have many terabytes worth of data before it's as cost effective as external/portable IDE hard drives. The cheapest is still DVD burning, however, there is a lot of manual labor involved.

Tape drives can be affordable

A brand new tape drive with acceptable speed and capacity is typically prohibitively expensive for home users and small businesses. However, they can be bought second hand. For example, for less than $100, I recently acquired a DLT 8000 drive (40 GB uncompressed data per tape) which connects to a SCSI bus. The drive is also portable, so that it could potentially be used to take backups (or other types of copies) at different locations (for instance at home and at work).

SCSI is by far the best external interface to use for transferring large quantities of data, although Ethernet (especially Gigabit) and Firewire may be acceptable. USB, on the other hand, is technically inferior and seems to have price (and thus availability) as the only argument in its favour. (Although SCSI cards are also very cheap these days, especially from eBay and similar sources.)

Interesting thread. Thought I

Interesting thread. Thought I would write a little blurb to update this information to 2009.

It seems that today, the most efficient and cost effective backup method happens to be SATA drives.

Although non-typical, we are starting to see home storage devices that carry 2TB or more of storage. In 2009, companies are starting to offer 4TB storage capacities in NAS devices designed for small business and home office use. (See Buffalo, NetGear, etc.)

In order to backup this amount of storage, an LTO4 is probably the most appropriate, with a capacity of 800GB/1600GB Compressed. Unfortunately, at the time of this writing, LTO4 tape drives cost around $1000 each, with LTO4 tape media costing as low as $40 each.

Enter the SATA Drive Dock. These drive docks are designed to interface with a bare SATA drive, without installation into an enclosure. The drive is seated into the dock much like a cartridge which then connects it to a host computer or server. Purchasing a USB 2.0 or eSATA SATA Drive Dock (roughly $30), and individual SATA disks (roughly 1.5TB for $130) has more appeal than a tape system from an initial investment standpoint.

There is higher degree of convenience, no maintenance on a tape drive, no cleaning, and backing up to a SATA disk over USB 2.0 or eSATA is roughly faster than using a tape drive. (480Mb/s and 3Gb/s respectively.)

There are also companies that offer storage cases for SATA drives (Search DriveBox).


Thanks a lot for the update.

Currently, I use two 1TB external drives connected by USB 2.0 to a Linux server. It works fine. I setup a script that uses dump to do incremental backups daily and full backups weekly. More details at setting up a hard disk USB 2.0 enclosure for backup under Linux. The only drawback is that there is no (easy) offsite backup, and it eats up a lot of CPU when it runs, because that is how USB is.

I have looked at the devices you mention briefly and once the server is upgraded to an eSATA motherboard, I am getting one of these NexStar eSATA drive dock.

One thing that has not been

One thing that has not been addressed at all is longevity. How long do you want your data to be retrievable?
Hard disks seize up in a year or two if they aren't kept spun up.
Optical media decay, with significant data loss in just a few years.
Tapes still last for decades with zero data loss, and the structure of the data on a tape makes it easier to recover than a hard drive. Some tape formats have built-in error correction, so that even when bits are lost, the data can still be re-constructed and recovered.

Data backup is like insurance: you decide how much catastrophe you are willing to accept, and purchase a policy [data backup system] that will cover that.

If losing your data would be a bummer, back up on DVDs.
If losing your data would cause serious headaches, back up on a RAID.
If losing your data would ruin you, back up on tape.

RAID is not backup ...

Having a RAID array is not backup as long as it is not versioned, not disconnected from the machine it is in, and not stored offsite.

Can't agree more...

We had a young tech think having a RAID was a preferable replacement for our tape library, until the RAID lost it's backplane and corrupted 7 of the 16 x 2TB drives in the RAID. Luckily we had the previous nights data backed up on tape. We lost a days worth of new files and modifications, after 8 hours of restoring the data it was far better then sending the whole RAID off to a drive recovery facility and paying them $72K to rebuild the data.