With ever increasing demands for data storage on magnetic disk and more efficient data distribution media, the need exists for alternative means of storage and distribution media. Recent developments in Compact Disc-Read Only Memory (CD-ROM) technology and changes to the ArcInfo data file naming structure have made CD-ROM a viable medium for archiving and distributing ArcInfo data sets. CD-ROM publishing systems make "once-off" CD production a very effective tool in any Geographic Information System (GIS). CD-ROM technology is also the perfect medium for solving data transfer problems associated with interoperability across different computer platforms. International Standards Organization (ISO) 9660 format with the new ArcInfo file naming structure allows ArcInfo data to be accessible from virtually all computer platforms using ArcInfo or ArcView-II.
Prior to the release of ArcInfo [1] version 7.0, putting ArcInfo data on CDs was difficult. The INFO directories required file names containing nine characters (i.e., arc001dat) and coverage file names were three alpha characters with no suffix. This posed a problem for some CD publishing software that would support only DOS "8.3" file naming conventions, known as "Level 1" CD format. The ISO 9660 standard CD format does not have "8.3" file naming limitations, but most CD writing software assumed this so that all platforms could read the CD, with DOS being the lowest common denominator. There were several workarounds to these problems, such as the use of Rock Ridge extensions, "Level 2" ISO 9660 format, or creation of a file system image to the CD, but these options have not been available on some of the less expensive CD publishing systems.
ArcInfo version 7.0 solved most of the CD format incompatibilities. INFO directories and geodata set directories converted to "8.3" file names (i.e., arc0001.dat). Geodata set directories added ".adf" to all file names (i.e., pat changed to pat.adf) except the log file name. However, to avoid having CD publishing software put a dot (".") after a workspace log file name or geodata set log file name, which makes the files unaccessible to ArcInfo, a "Level 2" ISO 9660 format must be made.
With the advent of ArcView-II, which operates under DOS/Windows, UNIX, and Macintosh operating systems, the ArcInfo version 7.0 file structure makes CD-ROM the perfect medium to distribute UNIX ArcInfo data for viewing and query for all platforms. Geodata sets need not conform to the DOS 8 character directory name limitation as long as the first 8 characters of all geodata sets in a single directory are unique. ArcView-II will truncate and encode these geodata sets so they can be accessed individually.
CD publishing systems for UNIX platforms have recently become competitive with those of PC/DOS and Macintosh platforms. At this writing (April 1995), a UNIX based CD-ROM writer with CD publishing software can be purchased for around $4000. USGS is using a UNIX based CD publishing system consisting of Gear CD publishing software [2] on a Sun UNIX platform with a Philips CDD522 CD-recorder [3] to produce "once-off" CDs for distribution and mastering of ArcInfo geodata sets and related software. Blank CDs cost about $12 to $13 each, making them less expensive than the popular Quarter Inch Cartridge (QIC) tapes. CDs will hold between 580 and 680 megabytes of data that, unlike data on magnetic tape, can be accessed "live", albeit slower than data on hard disk. This ability to access the data live makes CD-ROM an optimal medium for archiving data and programs. The life expectancy of CDs is rated at 50 to 100 years, whereas magnetic tapes should rarely be used past 10 to 20 years.
In the past, the use of CDs with dynamic data has been discouraged because they are "write-once". Read/write optical disks have been a prefered medium for storing dynamic data, but the costs are high at $200 per platter, and distribution can only be between like platforms. With CD costs as low as they are, dynamic data are no longer a problem, especially for storage on "once-off" CDs, and the CD provides the added benefit of a historical archive. Mass producing CDs has also become very inexpensive. Costs range from $10 to $12 per CD for 100 CDs and $3 to $4 per CD for 1,000 CDs. The more CDs that are produced, the lower will be the unit cost.
CD-ROM manufacturers will accept data on magnetic tape to be mass produced on CD-ROM. However, the data cannot be adequately tested until it is actually on a CD. One can attempt to reproduce the behavior of CDs by restricting write access on magnetic disk or by mounting a magnetic disk read-only, but this will not completely simulate the data and data access behavior because the system is not interpreting it through a CD driver. The ability to put data on a "once-off" CD for testing before sending it to be mass produced on CDs may save the user from reproducing costly errors. For ArcInfo data to be accessed on UNIX platforms it is often desirable to use UNIX variables on the CD so directory pathnames do not need to be hard coded. Again, this can be tested only on a CD itself.
In terms of the costs and effectiveness of different automated storage media such as CD-ROM, magnetic tape, read/write optical disks, etc., CD-ROM is by far the best choice for storing or distributing ArcInfo data, as well as most other types of data. From live access to platform independence, no other medium can make as strong an argument. Spatial data and programs are not the only features being written to CDs. As an example of commitment to using CD as a prefered medium, ArcInfo documentation is now exclusively released on CD, which not only is very cost effective, but also conserves resources.
[1] Use of trade names does not constitute endorsement by the U. S. Geological Survey
[2] Gear CD publishing software - Elektroson USA, 31 Waterloo Ave., 2nd Floor, Berwyn, PA 19312, (610) 617-0850
[3] Philips CDD522 CD-Recorder - Philips Consumer Electronics B.V.,Interactive Media Systems,P.O. Box 80002,5600 JB Eindhoven,The Netherlands,+31 40 7 35440