EMedia Professional - Standard Deviations, January 1997

Why CD-R Will Never Be a 680MB Floppy

Dana J. Parker

EMedia Professional, January 1997
Copyright © Online Inc.

Unsuspecting packet writers will soon discover that the packet-writing Promised Land they have long been awaiting will fall far short of its advanced billing.
At long last, packet writing is here. The logical file format, CD-UDF, has been decided on and endorsed by the Logical subcommittee of the OSTA CD-R committee, and support for the format will be included in the next update to Windows 95. So one could conclude that the CD-R disc stands on the edge of a new frontier, poised to realize its promise as a 680MB floppy.

But there is something the purveyors of CD-R packet-writing solutions aren't telling you. Even though the logical impediments to an incremental writing standard have been largely eliminated and the buffer-underrun blues have been shaken once and for all, there remains a major physical impediment built into the Orange Book specification that will curtail packet╩writing's usefulness. Unsuspecting packet writers will soon discover that the packet-writing Promised Land they have long been awaiting will fall far short of its advanced billing.

For instance, a conscientious CD-R user may decide to back up his or her hard drive to CD-R on a daily basis, using packet-writing technology. If the files updated daily amount to 100KB with the necessary overhead for modified directory structures, the data stored on the CD-R backup will grow by roughly 150KB every day. Faithfully backing up seven days a week, the user can expect to store about 55MB per year, and use the same backup disc for nearly twelve years before it is full. But on Day 101 of the first year, the user inserts the disc and discovers that the recorder refuses to write to the disc, reporting that the disc is full. With only 15MB stored to date on a 680MB disc, what gives?

PACKET WRITING'S FIRST PITFALL: THE PCA PROBLEM

The reason for this unpleasant discovery is found in a basic Orange Book structure called PCA (Power Calibration Area). This area is hidden from the user but nonetheless is intrinsic to every recording operation and physically limits the number of times a disc can be recorded. The PCA is located on every piece of CD-R media near the hub, at a negative time address. It is invisible to a CD-ROM drive or CD audio player, but it is the first place a CD recorder looks each time a user tries to write to the disc. The PCA consists of two structures, a Test Area and a Count Area. The Test Area is used to calibrate the power of the recording laser before every recording operation, and the Count Area is used to keep track of the available space in the Test Area.

A vast range of external forces can affect CD recording, from variations in ambient temperature and humidity to disparities between types of dye polymer and even between batches of dye polymer from the same manufacturer, to variations in the sensitivity of a disc's recording layer over its lifetime. The recording characteristics of the disc itself are defined within a range of values. Optimum Power Calibration (OPC) is intended to adjust the recording laser power to the optimum level, dependent on all of the variables involved.

Every time a disc is inserted in a CD recorder, the recorder first scans the Count Area of the PCA, which contains 100 numbered partitions, each partition one Absolute Time in Pregroove (ATIP) frame long. These frames correspond to partitions in the Test Area. After the power calibration is performed in the Test Area, one of these Count Area frames is recorded in sequence with random data. The recorder counts the number of unwritten ATIP frames to determine where to perform the next test operation. After 100 recording operations--in some recorders, after 100 insertions of a given disc--the Count Area becomes full, which means no more available addresses for the Test Area and no further power calibration. The recorder perceives the disc as full, so it can no longer record, even if the disc contains less than 10MB of data.

SEVERAL SOLUTIONS: CALIBRATION COMPROMISES

The OSJ (Orange Study Group of Japan), a group of CD-R media and drive manufacturer representatives, proposes to diminish--but not eliminate-- the "disc full" problem by redefining the PCA. Their approach would further subdivide the existing allocated space for each OPC and Count into 10 to 20 percent of what is currently used. This would increase the number of allowed power calibrations (and recording instances) to 500 or 1,000. Although this would be an improvement, it would only increase the effective capacity of the disc to 150MB for the extreme backup situation cited earlier; its useful life would be extended to two years and nine months. Granted, a seven-dollar disc used for two years of daily backups is a bargain at less than a penny a day, but 500MB is still a lot of space to waste.

Another option, which no one is seriously considering, would be to modify the space reserved for PCA to accommodate the worst-case scenario in packet writing--an instance in which a disc is written a packet at a time and removed after each packet is written. At one calibration and count area per packet, this would require 50,000 calibration areas, or an extra 50MB of disc space. 50MB is also a lot of space to waste, but it is the only way to ensure that packet-writing CD-R users will not encounter the "out of test space but not disc space" situation. Combined with the OSJ approach of using reduced space for each calibration, this would still waste 5MB of disc space. If a compromise figure of, say, 10,000 calibrations were used instead, only 1MB of space would be wasted. Unfortunately, using the "set aside" approach would waste the space no matter how many times the disc was recorded.

The most sensible and least wasteful approach would be to allow incremental allocation of power calibration space as needed. This would require a modification of new drives to allocate one packet's worth of space for calibration in the program area of the disc, leaving the "old" PCA area for use by older drives. Allowing one calibration to be performed per calibration packet would add at least 16KB (in variable length packets) or as much as 64KB (in fixed-length packets) per writing instance. If one calibration packet could be used for several calibrations, this overhead could be reduced, since each packet requires 14 sectors of run-in and run-out blocks.

But even this solution is not without its drawbacks. According to Andy Young, president of Young Minds and a member of the OSTA CD-R Logical subcommittee, OSTA is proposing a complementary approach which may reduce the need to recalibrate the laser power as often. A proposal now under consideration by OSTA's CD-W Logical Format Committee suggests that a field be provided within the CD-UDF file system structure which would allow information about the levels from the most recent laser power calibration to be recorded on the disc, along with a drive type identifier, drive serial number, and a date and time stamp.

"If the disc has been written in multiple drives," Young says, "the most recent calibration levels for each of the drives could be recorded. Thus the next time such a disc is inserted in a drive which it had been written in before, the software could recognize this and download the most recent calibration values." After writing the first packet, the recorder could check the number and type of any write errors, and perform no further calibration if the write was successful. On the other hand, Young says, "If excessive or significant errors are detected, the laser power could then be recalibrated and the first packet rewritten before continuing."

ALL OVER BUT THE SHOUTING: TOO LITTLE, TOO LATE

So now that we've identified the problem and proposed several solutions, what more do we need? The simple answer: an agreement on the method and someone to take responsibility for enacting it, and, ideally, the ability to go back in time. OSTA's position is that it suggests implementations but does not address changing the Orange Book or any other specification or standard. The OSJ, while it does address physical changes, only makes suggestions as well, but if every member acts on the suggestions, they are as good as a requirement. Even so, both organizations are loath to suggest such a major change in the Orange Book; redefining the number of test and count areas in the CPA--a media-only solution--is about as far as they are willing to go. With two years already spent awaiting a logical solution, there just isn't time to wait for a physical change.

This leaves us with CD-R packet writing as a mirage in the quest for the 680MB floppy. The window of opportunity for CD-R packet writing is long gone--at least two years late and several thousand OPCs short. That doesn't mean packet writing won't offer relief to those who envision a standardized, large-capacity, efficient optical storage solution for backups and internal use; it simply means that CD-Recordable won't be the medium that benefits from it. Perhaps CD-Erasable will.

Dana J. Parker is an independent consultant and writer and regular columnist for Standard Deviations. She is the co-author of CD-ROM Professional's CD-Recordable Handbook (Pemberton Press, 1996).

Home Page