Around the Bizauto web site:
BIZAUTO BULLETIN 97.2
Security You Can Bank On
Cyber-Soothsayer
'97
and
the news
Past
Bulletins
Who
is Business Automation?
Our
services
Consulting
projects
Expert
witness projects
Home
|
FEATURE ARTICLE:
SECURITY YOU CAN BANK ON
(THAT WON'T BREAK THE BANK)
This Bulletin, the eleventh
of a series, reviews the major aspects of computer security, from basic
data integrity safeguards to the newest advances in unauthorized access
protection. Past issues in this series have discussed system components,
printers, data storage systems, networking and data communications. Future
installments will cover business software and the use of the Internet for
business purposes.
Security is a critical issue for
every computer system, but as important as it is it still can't be looked
at in isolation. The tighter a system is "buttoned up", the harder it is
to access . . . even for authorized users. An important concept is that
security doesn't have to be perfect, it just has to be significantly more
costly to break than the value of the information gained by breaking it.
Once a system crosses this basic threshold, the value of investing in further
security disappears. That's why, in judging any security procedure, it's
important to weigh it against system usability. The truth is, even if there
were such a thing as perfect security (which there isn't), it's not likely
anyone would want it because it would be so cumbersome to use.
This Bulletin considers three
basic security issues: the security of data against destruction
or loss, the security of systems against outside penetration,
and how Internet connectivity affects security.
BACKUP: THE MOST
BASIC SECURITY
Data is the core of any business-oriented
system and keeping that data safe and accessible is the most basic principle
of computer security. Magnetic tape has always been the most common medium
for data backup, because of its relatively low cost. However, since today's
software often fills up huge amounts of data storage capacity, digital
audio tape (DAT) drives have become the most common backup medium for most
new business systems and networks. But now when even desktop computers
come with more disk capacity than older file servers even DAT drives, which
can store up to 16 gigabytes on a single tape (i.e., 16 billion characters,
or 2000 big-city phone books of data), may be too small. This is particularly
true in networks where the individual users' computers are being backed
up, as well as the servers. To meet this need, the tape drive manufacturers
have begun to deliver auto-changers that can do unattended backups to multiple
DAT-cartridges. The only drawback to these units is speed.
With today's technology, it can
take a couple of hours to fill each tape so that no more than a few could
be used in one night. This will change in the future, however . . . a new
tape-writing technique more than ten times faster than DAT should be coming
out soon.
Backup media, other than tape, are
also coming into use, including rewritable CD-ROM and optical disk. These
have the benefit of being faster, but they cost quite a bit more than tapes
and usually have much lower capacity.
With networks becoming more widely
used for "mission critical" business applications, the need for constant
"real time" backup has increased. This includes several ways of storing
data on multiple disk drives simultaneously (e.g., disk mirroring and RAID).
Old multi-drive systems forced users to shut down to replace a failed disk.
But the newest ones are "hot pluggable", which means they can be replaced
after a failure without either turning off the power or interrupting network
operation. Though these systems that regenerate themselves to their full
redundant state after the failed drive is replaced have been available
for several years, it's only within the last year that they've become common
in moderately priced servers.
It should be noted that making tape
backups is still important, even with a redundant storage system, for data
archiving, off-site storage (to protect against fire or other hazards)
and protection against the simultaneous failure of more than one drive.
LOG-ON AND DIAL-IN
SECURITY
Backing up protects against equipment
failures and other disasters, by far the most common causes of data loss
and corruption. But what it doesn't do is protect confidential data against
the threat of exposure inside or outside the organization. Small businesses
often ignore this risk, forgetting how sensitive things like payroll/personnel
records and customer lists can be. That can be a serious mistake.
Password usage has always been the
easiest and most common method for preventing unauthorized system access.
But passwords aren't prefect, particularly since users often choose passwords
that are common or easily-guessed. They also tend to become known around
an office, making them susceptible to unauthorized use by disgruntled current
or past employees. One solution for this is the use of software that prompts
users to change their passwords regularly.
Another good safety measure, particularly
in environments where computers may be left on in public areas is the use
of "screen-saver" software (which once actually saved screens from phosphor
"burnout", but aren't really needed for that any more). Today, screen savers
are most useful for their ability to obscure and lock unattended computers
so that passers-by can't use them to access confidential data.
Of course, some industries, such
as banking and defense, need more rigid security than passwording alone
can provide. There are two leading technologies for this kind of protection:
"biometric" verification and variable security code generation. Biometric
systems used to be prohibitively expensive but have recently become more
affordable. They use unique physical characteristics such as fingerprints
or retinal patterns to verify users' identities.
The variable security code technology
is only slightly less secure and it works for mobile, outside-the-office
users as well. With these systems, every user has a hand-held code generator
card, similar to a credit card size calculator. These cards display constantly
changing codes that must be entered by the user as part of his log-on sequence,
much like a password. The codes are unique for every card and are time-synchronized
with a master code verification unit that protects the system or network.
Thus, if one is lost or stolen, an administrator can remove it from the
authorized list, making it useless until it is back in the proper hands.
A final form of dial-in security
for outside workers is "dial-back", which now comes standard in many communications
software packages. This technique works by enforcing a two-step connection
process. When a call comes in, the calling computer identifies itself,
and the system responds by hanging up and calling back a predesignated
number for the identified caller. This prevents unauthorized access by
intruders calling from a different location. Dial back doesn't work for
travelers (because they don't have a fixed dial-back number) and can be
foiled by "phone hackers" (although that's very difficult), but
it is quite reliable and considerably less costly than code generators,
so it fills a useful niche.
E-MAIL AND THE INTERNET
Increasing use of the Internet has
created a new series of security problems, the most feared of which, transmitting
(i.e., e-mailing) credit card numbers, may be the least severe. The reason
for this is that few e-mail messages have any commercial value at all and,
of those that do, most can be quite adequately protected using readily
available encryption techniques (described below).
"Counterfeiting" of e-mail is potentially
a much greater risk than interception. Imagine the problems that could
be caused by forged e-mail instructions about where to send an order someone
has placed . . . that's just the beginning of the problem. With today's
Internet e-mail systems, it's extremely easy for someone to forge an e-mail
message with whatever "return address" they like. With only a little more
effort, a counterfeiter can route the bogus message through all the same
Internet "mail servers" as a genuine e-mail, making it very difficult to
spot as a fake. Some counter-measures are available to prevent e-mail forgery
(e.g., a way of putting authenticated "signatures" into an e-mail message)
but they're not widely used. The US Postal Service is developing "Electronic
Postmarking" that would authenticate the time/date stamp on e-mail
messages but, so far, that's as far as it will go. All the major online
services require a password to send e-mail, which makes them much more
secure for end-to-end communication than Internet e-mail. Thus, firms needing
that capability now should consider one of them for that purpose (particularly
Compuserve, because it has the most extensive international network).
Data encryption
(i.e., scrambling the messages) is the best way to transmit highly secure
information across the Internet. Several software packages now available
scramble messages so well that even the fastest computers would take days
to unscramble a few sentences (assuming the decryption "key" hasn't been
compromised). Until recently, it has taken several steps for both the sender
and receiver to use encryption, but some newly released software should
ease this task considerably. These techniques still face some regulatory
hurdles, but they will hopefully be eliminated soon.
The other big Internet-related security
issue is the risk that linking an internal corporate network to the Internet
could open it to penetration by industrial spies. This is a major concern
in highly competitive industries. The best way to handle this is, whenever
possible, to severely limit the ways in which the network is interfaced
to the net. If practical, the interface should be limited to e-mail only
(with a virus filter to screen incoming messages and message attachments).
It's also best to use a totally separate network possibly one maintained
by an outside supplier to host the company's World Wide Web pages. If these
pages need to include status information from the company's corporate database,
this information should come to the separate network via a secure one-way
data transmission line outside the Internet. When this isn't feasible,
there are systems called "firewalls" can provide reasonable levels of protection.
However, firewalls can be quite costly (although prices have come down
in recent months) and can require great technical expertise to implement
and maintain.
Click here to read the previous
article: Network "Talk"
Agree
or disagree? . . . Take it up with
the author.
|
