arrays, and CD-ROM juke boxes.
age media upgraded as needed. Move the data from
In this arena hard drives and disk arrays have the
obsolete media to current standards.
advantage of flexibility and speed, which is not neces-
Currently write/serve jukeboxes exist for both CD-
sarily as important as reliability. Most hard drives fail
ROM and DVD. This eliminates dependence on exter-
within 3 to 5 years. CD-ROMs have the advantage of
nal readers and allows batch conversion and renewal.
reliability with a sacrifice to speed and flexibility. If a
dataset must be changed for any reason, the CD must
Problem: Backup storage
be recut. Because CDs are read-only, they do not need
to be backed up, although if you are cutting a CD you
Backup storage is used to recover from accidental
generally cut more than one, which serves as a backup
data loss, unauthorized tampering, and disasters. Data
itself. Properly executed, either is a viable option. In
are important. Data are money. Loss of data is a loss
current technology, a large disk array properly backed
of productivity and money. Backup of data is the
up is generally easier to set up, administer, and use.
cornerstone of any good data storage methodology.
It provides the speed users want with the flexibility
Any decisions made regarding the types of data stor-
required by administrators. Large hard drives are cur-
age described above must take into consideration the
rently much cheaper than implementing large CD-
eventual need to back up and restore data. All sites
visited that performed regular backups used some
form of tape media.
Problem: Archival storage
Archival storage methods are used for datasets that
Tapes are reliable (within their accepted limita-
are no longer active. Frequently these are datasets that
tions), cheap, and easy to use and administer. The
have been superseded or outdated for some reason.
choice of media is a quickly changing target. Most
sites used a variety of 8-mm helical scan tapes that
have been an industry standard for several years. This
These datasets should be kept for historical rea-
medium is typically good for storage of up to 14 Gb of
sons and to fulfill community requests; instant access
is not necessary. Archival storage is generally met
The newer standard is Digital Linear Tape (DLT).
by tape media in one form or another or by CD-
This medium is more reliable, has a larger capacity,
ROMs. CD-ROM is the better choice for this type of
and is considerably faster given the correct circum-
storage in most cases, although it is initially more
stances. If no standard currently exists at your site,
then DTL should be the medium implemented for
Although this type of storage is easy to fulfill with
tape media, there is a frequently unnoticed problem
One important step in a successful backup strategy
associated with it. Due to the long period of time
that most sites failed to take was validation of back-
between requests for this data, the media on which
ups. Tape media fails, especially if used/rewritten
it is stored can easily exceed its intended lifetime, or
often. Tape writers fail. Networks fail. Backups made
become outdated, with no means to read it back onto
regularly should occasionally be checked for validity.
a useful medium for transmittal. Several of the sites
Can you restore data from the tape, and are they in
visited had wall racks of tapes that could only be read
the same condition as when they were written? Tapes
on one 1015-year-old computer. When that computer
should be retired before they reach their useful end
fails, the information stored on the tapes will likely
of life so that loss of data will not occur. These steps
will help to ensure that the data collected today will
Newer media like 8 mm, DAT, and DTL tapes are
be available when you want it tomorrow.
only rated for 3 to 5 years. After this time, the media
can fail or lose data without warning. CD-ROMS do
not currently have a known life span. Theoretically,
Underlying data storage concerns are concerns
they can retain their data indefinitely, and have thus
about the network architecture deployed at the site.
far proved reliable over periods of greater than 10
Many of the decisions to be made are based on speed
years. CD-ROMs have the added advantage of being
of delivery. The type of active online storage chosen,
considerably faster for retrieval of data than most
the medium for backups, and archiving are all tied to
forms of tape archiving.
the network upon which the data will flow. In addi-
If tape storage is decided upon, then the data on
tion, the choice of whether to keep files on individu-
the tapes should be regularly validated and the stor-
als' local hard drives or to store them centrally relies