In the Beginning, There Was Tape
August 30, 2012
By Brian Edminster
Applied Technologies
First in a series
In the beginning, there was tape. And if you’ve been around awhile, you remember it was on big reels about a foot across, was about a half-inch wide, and could have as much as 2400 feet of it on a reel. Yeah, they were heavy, too.
Data was recorded in parallel ‘tracks’ along the length of the tape. In this case nine of them, hence the name ‘9-track’ tape. At 800 bpi, that yielded a capacity of nearly 20Mb. Later technology allowed higher density, when 1600 bpi upped that capacity to about 120Mb. The last incarnation of 9-track was a whopping 6250 bpi — yielding nearly 1Gb of storage for a single reel of tape.
By comparison, anyone can get USB flash-drives that’ll hold 16Gb for $10 down at Walmart.
Very few, if any later model 3000’s (those that run MPE/iX vs. MPE/V) will even have a 9-track tape drive on them. And that’s a good thing. These 9-track tapes take up far too much physical storage space, and are far too slow to read and write. They might have been okay, back when disk drives were 50Mb, 120Mb, 404Mb, or even 570Mb (the capacities of the old HP 7920, 7925, 793x, and 7937 disk drives, respectively).
Unfortunately, a 2Gb drive is pretty much the smallest drive you’ll see on a 3000 these days, and larger drives are more common. This presents a problem: What do you do when it takes potentially dozens of tapes — and many hours — to do your daily backups?
Again, advancing technology comes to the rescue. Instead of further density increases on 9-track tapes, a variant of the digital cartridge tape called DDS (Digital Data Storage) came into common use for backups. For any new managers who haven’t seen one, a DDS cartridge is just a little smaller than a standard pack of cigarettes.
DDS drives and media also progressed in technology over time, growing in both speed and capacity (and generally, in reliability as well) as the table below shows.
As it turns out, transfer speed (how fast the drive can read/write data) also scaled upwards as capacity grew — so that even if you didn’t need the capacity, there’s often a speed advantage to the larger capacity drives. For any system that needs a new or replacement DDS drive, I’d never recommend anything less than a DDS3, unless there was an absolute requirement to write DDS1 tapes.
Also note that once you get to DDS3 and beyond, you’ll likely need a dedicated SCSI channel for the drive, in order to take advantage of the drive’s speed. If you cannot keep the tape drive’s buffers fed, it has to stop and re-position, which slows it down significantly. Note that this can occur while reading tapes as well.
If you have sufficient CPU capacity available, software compression can also speed up reading/writing of backups by reducing the amount of data sent to and from the drive. I’ve also heard from people I trust that you are unlikely to experience the difference in backup time which is implied by read/write speed difference between DDS4 and DDS5 — at least not when attached to a 3000.
As I mentioned in an article covering VSTORE on the NewsWire’s blog, DDS drives slowly lose alignment as they wear. They have a remarkable tolerance for this, all things considered (especially DDS3 and subsequent drives). It’s still something to be on guard for, especially if you’re relying on your backups to be readable when the chips are down.
Regarding media reliability: various sources give differing answers, but as a general guideline DDS tapes will take about 100 passes or so before they wear out. If you use the same tape once a week, expect to get about two years life out of it, before it wears out. You could get more uses, but the longer you go, the farther you’ll get into borrowed time territory. Don’t expect a catastrophic failure, but you’ll have to clean the drive more often. And face possible write failures and/or VSTORE failures.
Failure is not something you want to tempt by scrimping with your backup tapes. Buy the best quality you can, and replace them yearly. Like doing backups in the first place, it’s cheap insurance. Would a rock climber wait until his ropes break before replacing them? I know I wouldn’t.
Next time: New media for a new millennium