Altirium logo
Home Resources Articles Data Recovery Trouble With Tribology?

Trouble With Tribology?

Tape Data Recovery, my specialist subject since 1984, is a broad subject and actually relates to more than just tapes suffering from wear and tear. There are the perennial, publicity friendly, tales of apocalypse (fire and water damage) that make great reading but are (thankfully) relevant to very few of us. The majority of Tape Data Recovery work is required as the result of drive failure, media failure or human failure and these tend to happen during the normal course of events - and without a horseman in sight.

Recovering data from tapes is not a single subject. The action required for recovering data from a DLT or LTO that has been partially overwritten is radically different from that for AIT or DAT. The same is true when media problems cause a restoration failure, the actual recording technique used by the drive determines the action required. Then there is the application that wrote the data to the tape. Recovery work from a BackupExec or ARCserve backup is radically different from that required for a complex Tivoli, Networker or NetBackup recovery.

Under most circumstances the only differences between types of tape are performance, capacity and price. DLT and LTO which move tape at high speed and record multiple tracks serially along the tape, are operationally identical to DAT, AIT and VXA which record diagonal tracks along the length of the tape in one direction only. However, when low level action is required (as with data recovery) then the precise understanding of the technique of recording is key to the recovery methods required.

Disk round, tape straight?

This is general information about tape and I don't make any claims that it is 100% correct in all instances, there is a wide range of difference between drives and there have been many of changes over the years, and to cover everything would take an encyclopaedia. It is correct in most instances though.

There is much more information written about hard disks, and many people profess a great understanding of the principles of disk storage and data recovery from disk. Tape and tape recovery, however, leave many people in a state of trepidation. Tape has a reputation for being somewhat arcane, Explorer won't show you the contents of a tape, in most instances you cannot readily copy files to and from a tape, and many data recovery companies don't recover data from tape.

The differences between tape and disk are fairly irrelevant most of the time. A disk spins all of the time, a tape drive only moves tape when it needs to access data. Disks store data on rigid, circular platters. Tapes store data on long narrow coiled up flexible material. Another fundamental difference is that you don't usually take the platters out of a disk, but a tape drive is designed so that when one tape fills up you can continue on another one.

Both tape and disk store data magnetically though, and really the main differences come down to logistics and how storage methods are implemented. A disk, for example, can give you access to any sector on its surface in a very small fraction of a second. The time to get any sector is roughly the same. Tape, on the other hand, is long and narrow so to recover data from the current position along the media might be instant but to recover data from the other end of a tape could take minutes.

Disk is also a non-contact technology, if the heads hit the platter then it is game over in many respects. With a tape, the media is in contact with the read/write heads at all times when data is being accessed (for DLT, LTO and similar drives the media is in contact whenever the media is loaded in the drive, helical scan drives such as DAT and AIT to unload the tape from the heads when the drive has been idle for a while).

So, tape could be used like a disk but it would not be a very sensible way of using it as it would take too long and would wear out. Tape, therefore, has developed as a sequential archival medium, and sometimes as an intermediate "near-line" storage. Back in the dark ages, disks being expensive and unreliable, tapes were used to store data that could be requested at any time but where the access time was not a major issue. On a mainframe you could open a file then go and get a coffee whilst a computer operator somewhere found the right tape, mounted it on the drive and recovered the data that you needed. This approach is still taken in some data centres but it is not the primary use of tape storage.

One major difference between tape and disk is one of which you would probably not enjoy a demonstration. Take a DLT or LTO Ultrium tape in one hand, and a hard disk in the other. Raise them both to about 6 feet from the ground and let go. Once they have landed, see which one still yields data. I don't recommend trying this, (we do recommend that dropped tapes be retired from use), but 90%+ of the time the disk will no longer work but you will be able to recover data from the tape. This does not mean that one is better than the other, they just serve a different purpose.

Tape is a sequential storage medium, disk is a random access medium. Disk is suitable for rapid access to any position on its surface (so perfect for the demands of real time computer activities), tape is lower cost more durable and better suited to longer term archival storage and recovery.

Other tape features include

Physical Beginning of Tape (PBOT)/Physical End of Tape (PEOT)

These are as they say the physical start and end of the tape, not necessarily where the data starts and ends. There are tape cartridges that start recording half-way along the tape, others record along the tape then back again several times and actually finish recording back at the physical beginning of the tape.

Block Positioning

When recovering data from a disk the command issued contains the disk sector number. To recover a block (a chunk of data on a disk is named a "sector" on a tape it is named a "block") a read command is issued and the next block is read. To recover a specific block an additional command must be issued to position the media before the read is issued.

File Marks

Tapes encode non-data entities that you won't find on disk. These being File Marks (aka Tape Marks) and Set Marks. The purpose of these is to act as dividers that can be spaced to in order to facilitate the recovery of a selected range of data blocks. Backup sets are delimited by File Marks, on IBM AS/400 backups the object headers, data trailers are delimited by File Marks. To an extent these are a throw-back to the early 7 and 9 track open reel tapes where there was no concept of block numbering, and so the only way to get to the next object without reading and throwing away data was to skip to the next file mark. With some tape drives in the early 1990s it was only practicable to append data after a file mark so one had to be written.

Block sizes/inter-block gaps

Disks in most instances are formatted with sectors each of 512 bytes. It is practicable to format them with different sized sectors, on the AS/400 520 and 522 byte sectors are not uncommon, but all sectors will be the identically sized.

Tapes can be used as either fixed block or variable block size media, the block sizes can go as high as 16MB (this is theoretical, there are 3 bytes available for the block size specification, but try and get Windows to give you 16MB of VDMA memory and you will find that 512KB is the biggest you can probably manage). We have recovered data from tapes written in large UNIX systems where the data block size is over 1MB but this is rare.

There is no good reason for using massive data block sizes any more. On the open reel tapes of the 1980s there was a physical gap between blocks and so the fewer blocks written, the fewer the gaps and so the greater the capacity of the tape. With modern tapes such as LTO and DLT there are no gaps and block size is purely the amount of data transferred in a lump so writing 1MB blocks has no benefit and could make the recovery of data on another system rather problematic.

[In case anyone out there has read the specifications for the early Exabyte 8200 drives and some of the SLR drives up to QIC525, yes these did encode data in fixed chunks whatever you wrote to them. The Exabyte 8200 used a 1KB storage unit so if you wrote 1025 byte blocks to the tape each was stored as a chunk of 2048 bytes and you halved the capacity of your tape. You can rest assured that this is no longer an issue with tape drives.]

End of Data

End of Recorded Medium is not a concept found with hard disks. A hard disk as an amount of space available and all is accessible for use, it already contains data even if that is just some type of format pattern. Tape, has an amount of tape available but reading is only permitted up to the furthest point where data has been written. You write 20MB to a 300GB disk, you can still read 300GB of data from the disk. You write 20MB to a Tape, then all you can read is 20MB from that tape. It does not matter if there were 60GB already written to the tape, if you write 20MB at the start, then that is what you can recover, the rest is no longer accessible without resorting to data recovery.

There are other features of tape, such as partitioning, that differ from disk. Partitioning on a tape is performed by the drive rather than the operating system and makes a single tape have the characteristics of 2 or more tapes. It is not greatly used.

By far the major thing with Tapes that sets them apart from Disk in most peoples lives is the way they are used and the multitude (though greatly diminished from 10 years ago) of different storage formats used. Plug a USB disk containing an NTFS partition into your Windows system and you can use explorer to access the files. Put an LTO or DLT in your drive and if it was not written using a piece of backup software that you have installed then your chances of even identifying what is on it are remote.

[Before anyone gets picky, yes BackupExec and ARCserve have a degree of interchangeability and might each recognise and recover data from a tape written by the other].

There are also issues particular to each format. We have recovered data in many instances where Tivoli Storage Manager tapes have been expired, meaning the database that stores the information about the data has marked tapes as no longer in use. The tapes contain data but from what we have been told it is difficult to get Tivoli to scan the tapes and accept the contents.

Cassette versus Cartridge

Although generally referred to as tape cartridges or data cartridges, there is a split between cartridges and cassettes. Cassette type media, where the tape runs from a supply reel to a take up reel within the same casing (DAT, AIT, Exabyte, VXA, 9940, 3570, SLR...), and cartridge type media where there is a single supply reel in the casing and the tape is taken up and wrapped around a take up reel that is build into the drive (LTO, DLT, SDLT, 3480, 3590, 3592 ...).

Then there are two basic methodologies of recording that are followed, these being helical scan which is like the VCR you used to have at home, and linear recording techniques as used in drives from 3480 through to DLT and LTO.

There is a third format that is somewhere between linear recording and actually treating a tape like a disk. This began with mini data cartridges of the DC2000 ilk, well some way earlier with DC100 on HP pre-PC systems, but DC2000 is the starting point for the current generation.

Helical Scan

Helical Scan is the recording technique used by Exabyte, DAT, AIT, VXA and some other drives. The Exabyte 8mm drives derived from the Sony Video 8 camcorder whereas 4mm DAT developed from Digital Audio Tape that became popular for a while in the audio market. VXA and AIT are also 8mm technologies.

The tape is wound from one reel to another within a cassette, but in the middle it is drawn inside the drive mechanism and wrapped around a cylindrical head assembly. The cylinder spins rapidly as the tape moves past slowly. The cylinder is set at an angle causing the read/write heads to "describe a portion of a helix" across the tape. So the tape moves past and diagonal data tracks are recorded across it.

With these drives (if there is no partitioning) the recording starts at PBOT and ends at PEOT.

Linear Recording

Linear recording has been with us for a long time, since well before helical scan reared its head. The first open reel tapes used a linear technique, as to the latest LTO and DLT drives.

It is what it says, recording in a straight line along the length of the tape. Where differences occur is in the implementation. The early open reel tape drives wrote data as 7 or 9 simultaneously written parallel tracks along the length of the tape from PBOT to PEOT. The early SLR (then known as QIC) drives recorded on track at a time, but since this left most of the width to the tape media unused they would get to PEOT, then reposition the recording head and record another track from PEOT to PBOT. QIC11 drives recorded 4 tracks, then as the technology improved the number of tracks was increased to 9, then 15 and so on. the latest SLR drives record 2 tracks at a time, but follow the same method. This method is known as "Serpentine", and most drives now will end up at PBOT if writing a full tape.

The evolution from the open reel tapes was to the single reel tape cartridge, 3480, 3590 and so on. The 3480 increased the number of tracks written simultaneously to 18, then the 3490e managed to fit the 18 tracks onto half of the width of the tape and then fit a further 18 tracks from PEOT to PBOT.

DLT evolved from the DEC TK50 and TK70 drives and followed an SLR serpentine type technique of multiple single of pairs of tracks, as did LTO.

Drives such as IBM's Magstar 3570 and STK's 9840 (which use cassette type media) further complicated matters by adding mid-point loading to the equation. These were designed for near-line storage, for relatively high speed recovery of files, so it was reasoned that if they loaded in the middle they would never have to traverse more than half of the length of the tape to recover the required data.

The 'others' - almost tape, but a bit like a disk

In the 1980s when tapes have QIC, PERTEC, SCSI and other quite powerful interfaces the notion of attaching a tape drive to the diskette controller with in a PC came to the fore. As anyone who had done any extent of transferring data using diskettes will probably confirm, using the diskette controller to run a tape drive is a scary idea, but it was done.

DC2000 was the start point for drives such as Travan, Ditto, Onstream and a few other that came and went. Their advantage was that an expensive controller was not needed within the PC system, a major downside was that you needed to format your own tapes which took a long time and had you could end up with a badly formatted tape that would never be read again. (many drive manufacturers started selling controllers to get around these problems and to add features such as data compression in hardware).

Manufacturers of media started selling pre-formatted media, the DC2120, MC3010 and so on. The drive market was split in to two between Irwin who were dominant and the rest led by Colorado Memory Systems. Each supplied systems that were not compatible with the other's. Irwin fell by the wayside for reasons of commercial success, or lack thereof and the market was left with Colorado, Archive, Wangtek and a few others fighting it out for this end of the market.

The rise of CD-R in the early 1990's hit the market and forced it to try and develop. DC2120 cartridges, with 2:1 compression were superseded by MC3020 cartridges heading up to 400MB. At this point fights for domination really broke out with Sony producing QIC-WIDE with used 8mm rather than 1.4” tape in the cartridge to gain capacity, and then Imation launching Travan which used 8mm tape and changed the shape of the cartridge to allow longer tapes. QIC-extra appeared and then vanished, this was a way of having longer tapes that probably worked around the Travan patent.

Travan survives to this day though as IDE/SCSI /USB drives and not a diskette controller in site. Iomega Ditto and Onstream ADR developed from the QIC-Extra format but have since vanished.

As stated earlier this is not a technical specification for tape drives, it is a broad and general description of tape and some of tape's features and to reflect the diversity of tape storage formats that we encounter daily in our tape data recovery business. The manufacturers supply some quite comprehensive material on the drives they supply which is well worth a read if you intend to use one of their tape drives in anger.

Last Updated (Wednesday, 14 October 2015 14:08)

QMD 9001 logo
ISO9001:2008 Certified