Idling the disk (spin-down) also greatly enhances the lifetime of the mechanism. Instead of replacing the drive after 3 years of hard use, it gets replaced every 7 to 10 years (like tape).

Storage density is still an open issue but the relentless march of technology is solving that too. So for the next 5 years a company may need to allocate more physical space to hold the disks, chassis, controllers, etc. but over time it will shrink so the same amount of data can be stored in a given rackspace as with tape.

The following method developed by my former business partner, Dianne McAdam, would provide you with a rough estimate:

First, calculate the amount of storage required now and for the immediate future…say 3, 5 or 7 years for the sake of comparison. This should take into consideration anticipated data growth rates, as well as realistic utilization rates for disk and tape.

Next, using the figures from part one, calculate the hardware/software costs over that period. At a minimum, for tape that means the library and/or drive and cartridge costs. For disk, it means the drives, expansion units, controllers and racks.

Then add in electrical consumption (power and cooling) and floor space costs allowing for average annual increases. Obviously the overall environmental cost will vary depending on your geographic area.

That’ll give you a rough comparison. For a more thorough comparison, Dianne recommends including:

• Post-warranty maintenance costs
• Technology refresh costs (i.e., some companies replace disk every 2-3 years and tape drives every 5-7 years.).
• The cost of phasing in equipment purchases (i.e., it is not necessary to purchase all equipment in the first year. Not all of the disk controllers or tape drives and cartridges would need to be purchased the first year.)
• The cost associated with backing up data to local or remote tape or disk. This exercise assumes that only one copy of backup data exists and that additional backup copies are not stored at remote locations for disaster-recovery purposes.
• The additional cost of mirroring disk (RAID-1). Mirroring disk doubles the cost of the disk solution. Other forms of RAID protection, such as RAID-5, can provide protection against drive failures and require some additional disk capacity without doubling the cost of storage.

Using real-world current figures, she ran the numbers in late 2005 and found that tape came in at an amazing 11% the cost of an equivalent disk solution. When she presented her findings to end-users at a user group meeting in the fall they found her numbers to be, are you ready for this? - conservative.

She also ran the numbers for a hybrid tape/disk solution which offered the best of both worlds at a price that fell predictably between the cost of tape-only and disk-only solutions.

In a nutshell, if low cost is the highest priority, tape is still the logical choice. However, if quick access is more important than the overall cost of the solution, disk may be the better choice.

You’re welcome to take a look at the original analysis - send me an email. Heck, you can post the PDF on your site if you think your readers would find it useful. Though they should update the numbers with current values.

Kind regards,

Joe

The disk drives have the same basic design from the 50’s, only optimized. They have moving parts!
All other computer technologies moved ahead, a lot ahead… the hard disk drives are stuck in the past.

Perpendicular recording my ^&$… we need something _NEW_, faster, cheaper, no moving parts, smaller and when it dies, it becomes read only!

I would be a very happy man when the disks are replaced with flash (or other) type of memory. I can’t stand them anymore.

If somebody ask me, what is my biggest problem in IT at the moment - unreliable storage! As simple as that.

Where I work, out of 100 IT problems probably 90 are ‘dead hard drive’ or related issue. ~60% are using laptops working from home. The current disks storage technology is pathetic.

God bless the guys from Sun for ZFS :), it is about time for some real innovations in the way we store data. More is needed…

I wonder why we HAVE a storage management software market. I mean, if storage weren’t an expensive resource, why would anyone want to manage it? And why is it expensive? Partly growth of demand, yet raw storage is about 20,000 times cheaper than it was 25 years ago. So why is there so much more storage management software? My basic take is that our storage paradigm is so badly broken that we need all this stuff to keep the wheels from falling off. Which is why I like to look at new takes on the old problems, like ZFS and GFS.

Joseph,

You are correct of course. Yet IMHO the basic argument for putting up with tape is that it is significantly cheaper than disk. What sparked my comments is that Tandberg, a tape drive manufacturer, has finally stated the obvious: at the low-end their RDX disk backup is cheaper than tape. And it has all the advantages of disks: speed, random access, better shock resistance, wider environmental specs, etc.

Are there really environmental and management advantages to tape that you don’t get with disk, as your comment seems to imply? I’d love to have them sketched out.

Robin

I’m no rah-rah tape-vs-disk fanatic. I believe each media has its place, but if you’re going to make a legitimate, credible comparison, you MUST include the environmental and management costs.