Optical jukeboxes

Optical technologies
have cut deeply into the M-O jukebox market.


The measure of a jukebox varies with type of data and how it is used
Difference between
M-O and CD-ROM is in the details

Magneto-optical disk jukeboxes, once the darlings of big-time mass storage, have fallen
on hard times. Network managers in search of large-capacity data storage are instead
turning in droves to tape libraries and CD-ROM jukeboxes. But for some storage
requirements, M-O jukeboxes may still be the best choice.

M-O jukeboxes are the product of proven, scalable technology. M-O has advantages over
other storage options:

M-O is a little cheaper in large-capacity applications.

But there are downsides, too:

Although single M-O drives are not as cheap or portable as CD-ROM drives, single-disk
drives are available in compatible formats for use with PCs. That makes the data on

M-O disks portable and not accessible only through a giant and expensive jukebox.

At the low end of storage capacity, the cost comparison is very one-sided. A CD-ROM
changer costs as little as $250 and can carry 3.2G (five disks, each with 640M of data).

But when you need access to terabytes (1,000G plus) of data, you must turn either to
tape jukeboxes and their inherent problems—nonrandom access and vulnerability of the
media to magnetic field surges such as the EMF pulses from weapons—or to large-scale
optical storage.

For some LAN managers, it means using a 2.6G M-O disk jukebox to give users online
access to as many as 1,000 disks in a single cabinet. Users treat the M-O jukebox as one
gigantic hard disk, not just to retrieve data but for interactive editing of files, with
no concern about special formatting or the authoring process.

M-O technology is in no danger of disappearing. Many federal agencies have terabytes of
data stored on M-O media, and as the need for access expands, they will need to both add
jukeboxes and replace older units.

For very large data sets, robotic multitape libraries are an alternative to M-O
jukeboxes. But tape is only good for about five years of reliable storage without
rerecording, and M-O is nearly archival in nature, with an estimated life of about 50
years, comparable to that of CD-ROMs.

Unlike M-O, which is random-access, tape is a serial-access medium, sometimes making
retrieval of a specific file a time-consuming process. The media is unsuitable for most
online work in which users need to read and write data at random.

Older M-O technology—some still on the market and some found in legacy
systems—uses a combination of magnetic fields and laser beams to record and read

To store information, each sector must first be erased; all data bits must be reset to
zero before new data can be written to the sector.

To do so, the process subjects the area to a strong magnetic field in one direction. A
laser beam then heats the recording material, and the whole segment is changed to the same
magnetic orientation. The strong field used cannot be turned on and off rapidly, as can
the much weaker field used to directly overwrite a hard disk.

Rewriting is done in a similar two-step process. Data is first erased, then new data is
written. To record, a reversed field is imposed and only the nonzero bits are heated.

To read data from the disk, a laser beam is focused on the recording layer, which
reflects light differently, depending on its magnetic orientation.

Newer light intensity modulation direct-overwrite (LIMDOW) media and higher RPM drives
have made small format M-O drives faster by allowing one-pass overwrite capability and
improving average access times on read operations.

LIMDOW uses a dual-layer recording medium and can switch from read to write in
nanoseconds by modulating the laser beam instead of switching a powerful magnetic field
off and on. The drives are read and write-compatible with non-DOW ISO-standard 650M, 1.3G
and 2.6G MO disks as well as WORM media.

Stability is a strong point for M-O media. Unlike fragile, older 12-inch M-O drives
common in mainframe installations, the current 51'4-inch M-O disks use strong
polycarbonate plastic to protect the read-write data layer and can be removed safely from
the library.

Once stored on a disk, the information is archival. Exposing removed disks to either
strong light or a strong magnetic field won’t corrupt the data—only a
combination of the two will remove the information stored on a non-LIMDOW M-O disk.

Old drives can’t use the LIMDOW media, but the LIMDOW drives are billed as
compatible with old M-O and WORM disks. Drive upgrades may be the best option for users
who want to improve performance while keeping the ability to access existing archived

LIMDOW drives are available for several jukeboxes, including those from Plasmon Data
Inc. of Minneapolis, but check with vendors for details about compatibility and
availability if you’re considering upgrading an older jukebox with new drives.

Optical drives, including M-O, WORM and CD-ROM varieties, have advantages over hard

WORM optical technology lets you write data only to a blank disk and, although you can
continue writing data in various sessions until the disk is full, you can’t overwrite
data. Software switches on the disk and in the access drivers can be used to effectively
erase data by making the data unreadable, or you can update files and block off older
versions, but the disk itself just keeps filling up.

The disks can be used in the same optical jukeboxes as M-O disks.

Since the introduction CD-RW drives, other optical technologies have cut deeply into
the M-O jukebox market. A 650M CD-R disk costs less than $1, and even CD-RW’s disk
price is falling below $25 each.

CD-RW media cost about $20 per 650M disk in quantity. Although not as economical as the
$60 per 2.6G cost of storing data on standard 51'4-inch M-O media, the price isn’t
outrageous, especially when you consider the cost differential between 51'4-inch WORM and
CD-R write-once media.

Because the advantages cited for M-O disks also apply to CD-format media and drives,

M-O jukebox companies now find it difficult to compete. A visitor to one major M-O
jukebox vendor’s Web site finds a sad commentary on that difficulty. Except for a
single mention in passing, the company ignores CD-RW technology. It instead vaunts its
product’s superiority to CD technologies by saying that CD-ROM and CD-R are not
erasable. It then goes on to build a chart comparing the technologies but omitting CD-RW.

The chart serves best to point up that, unless you have a closet full of WORM or M-O
media, the older technology has trouble competing with CD jukeboxes.

This is not to say there are no advantages M-O enjoys over CD-RW. Your operating system
treats M-O drives as just another hard drive, while CD-RW disks usually require special
authoring software.

The story, however, gets grimmer with the addition of DVD. A chart on that same vendor
site comparing M-O and DVD products carries a fourth-quarter 1996 date.

Because they store more than the current CD maximum of 640M on a single disk, DVDs will
deal another blow to M-O. When it is widely available, rewritable DVD will likely drive
the last nail in the coffin of M-O jukeboxes.

Other media options include a 12-inch glass disk standard and Eastman-Kodak Co.’s
non-standard aluminum 14-inch WORM disk. These are best suited to mainframe storage
systems, possibly for a limited time. Kodak is considering dropping its proprietary format
drives and will make the medium only through the beginning of 2002. Agencies using the
Kodak system should contact their representative for further information. The latest
reports I could find did not indicate that Kodak had found a company to take over
manufacture of the disks.

Philips LMS may well be the only remaining company in the large WORM mainframe jukebox
market and the only supplier of the 12-inch optical drives. To find out about the 12-inch
Infinity optical jukeboxes, contact Philips Laser Magnetic Storage via FileTek Inc. at
9400 Key West Ave., Rockville, Md. 20850, tel. 301-517-1709; or check their Web site at http://www.filetek.com.

Hitachi Ltd. dropped out of the 12-inch drive market several years ago, and Sony has
announced it will concentrate on the 5'-inch drive market.

A major incentive in the move to the small media and new technologies has been cost. A
51'4-inch, 2.6G M-O disk costs about $60, or 2.5 cents per megabyte, but the drives often
cost in the $50,000 to $100,000 range.

Based on the number of disks you can store in a jukebox, that raises the effective per
disk cost by $100 to $200. A rough estimate would look like this: $160 to $260 per 2.6G
M-O disk, including drive and medium. The per-megabyte cost goes from 2.5 cents to 6.4 or
even 10.4 cents.

If some of your data is archived offline, you might have 500 disks online at 6.4 cents
per megabyte and another 5,000 disks offline at the far lower 2.5 cents per megabyte,
making M-O much more cost-effective.

Compare media costs:

The bottom line is that for a relatively small-capacity storage device it is far
cheaper to use a high-performance CD tower or even daisychained drives with one disk per
drive than to buy an M-O jukebox.

A redundant array of independent disks, or simply using many massive individual hard
drives daisychained in a nonredundant configuration, are also price-competitive with M-O

When you calculate which storage configuration will work best, you must add to media
and drive costs the costs of installation and maintenance. The cheap solution might end up
pricier than you think.

Measuring and comparing the performance of magneto-optical, CD-ROM, or write-once,
read-many jukeboxes isn’t as simple as it might seem.

Performance depends on the type of data stored and how the data is retrieved.

If your data consists of large files with few users frequently retrieving and saving
data to the same disk, all you need to consider is average access times and maximum
sustained data transfer rates.

But few jukebox applications are used that way. The average installation will involve
access of many disks by many users over a network.

While one user downloads a file, a queue of other users forms to get requested files.
If the first user is working with a file for a few seconds, then needs to rewrite a
modified version, he may be bumped to the bottom of the queue. He must then wait for the
disk to be reloaded to access another file.

Caching and disk management software can eliminate the wait if the file is just being
recorded and the user is moving on to a file on another disk. But there would likely still
be a wait while the other disk loads.

To determine performance under heavy use, consider how long it takes to access a new
file from the time the command is sent to the jukebox.

For the average jukebox with heavy traffic, this is how the action breaks down:

1.  Receive the file request.

2.  Determine which disk contains the file.

3.  Remove and file the disk currently loaded.

4.  Retrieve and load the new disk.

5.  Spin the new disk up to operating speed.

6.  Apply the average access time.

7.  Download the file.

8.  Ready for the next file request.

Calculate the time to deliver data by adding the time for Steps 1 through 8.

Performance is further affected by the number of drives available and the number of
robot arms that can be used to move disks simultaneously.

Other factors, such as the number of simultaneous requests, the percentage of times
that such requests call for data on the currently loaded disk, and even network
performance depend on the installation and can only be roughly estimated from the
published data.

Almost anything said in favor of magneto-optical disk drive jukeboxes can also be said
about the cheaper CD-ROM jukeboxes.

To decide between the two, you must look at the few differences, such as proven
scalability, performance and the amount and nature of data to be stored:

John McCormick, a free-lance writer and computer consultant, has been working with
computers since the early 1960s.


  • Records management: Look beyond the NARA mandates

    Pandemic tests electronic records management

    Between the rush enable more virtual collaboration, stalled digitization of archived records and managing records that reside in datasets, records management executives are sorting through new challenges.

  • boy learning at home (Travelpixs/Shutterstock.com)

    Tucson’s community wireless bridges the digital divide

    The city built cell sites at government-owned facilities such as fire departments and libraries that were already connected to Tucson’s existing fiber backbone.

Stay Connected