Norton 2000's our choice of PC readiness packages

TEST DRIVE

GCN Lab Product Table

There are 355 days left until 2000. As you scurry to ready the last holdout
systems—often including those pesky standalone PCs—here is a comparative review
of seven PC test tools.

Atop the heap is Symantec Corp.’s Norton 2000, which earned the Reviewer’s
Choice designation from the GCN Lab.

PC date issues occur in five layers: Layer 1—hardware, Layer 2—BIOS, Layer
3—operating system, Layer 4—applications, and Layer 5—data.

The BIOS layer works in tandem with the hardware layer—the real-time clock and
CMOS memory, which have changed little in the last 15 years.

The clock ticks off month, day, year, hour, minute and second; the CMOS statically
holds the century data as two digits. Because the clock and the century data are separate,
some utilities on the market might incorrectly flag a system’s clock as failing.

Ignore any Layer 1 tests that raise this red herring. Simply determine whether the BIOS
can correctly update the CMOS entry from 19 to 20 [GCN, Dec. 14, 1998, Page 1]. If so, layers 1 and 2 will likely do fine.

A good test tool must verify that the 2000 rollover occurs under all circumstances for
a variety of dates. At a bare minimum, it should verify that the BIOS and hardware
together correctly advance from Dec. 31, 1999, to Jan. 1, 2000, whether the system is on
or off. All seven products correctly checked this basic fact.

The test tool also ought to be able to check leap year dates during 2000, rolling from
Feb. 28 to Feb. 29 and then to March 1 without an erroneous Feb. 30 date popping up.

Network Associates Inc.’s McAfee 2000 ToolBox and Norton 2000 both did this
correctly.

Greenwich Mean Time-UTA L.C.’s Check 2000 PC and Check 2000 PC Deluxe, SecureNet
Technologies Inc.’s Y2K Test&Fix and ProveIT 2000’s package failed to check
the third leap year date.

I could not tell whether TouchStone Software Corp.’s CheckIt 98 tested the third
date.

Because 2001 is not a leap year, a good test tool also ought to confirm that the date
advances properly from Feb. 28 to March 1, 2001.

All the tested packages have patches for computers that fail the BIOS test. The patches
appeared to work on all the lab’s equipment.

Any third-party patch should be a last resort, however.

Most computer makers’ Web sites post BIOS updates for Pentium and later PCs. Use
makers’ BIOS updates rather than third-party patches wherever possible.

If a failing system has a 486 or older processor, check the manufacturer’s Web
site to see if a patch has been tested for the brand. Most patches are free.

The free YMark2000 utility from National Software Testing Laboratories Inc. of
Conshohocken, Pa., downloadable from http://www.nstl.com,
will test layers 1 and 2 for readiness, but it does not offer patches or BIOS updates.

Layers 3 through 5 are not as crucial as the BIOS and hardware layers, although it is
worthwhile buying Norton 2000 or one of the other tools for any PC with vital systems or
data.

The six packages that could check the operating system in Layer 3 turned out to be
inadequate. A good diagnostic tool ought to examine whether all appropriate OS patches
available from Microsoft Corp. have been applied.

Instead, the utilities merely checked to see whether the four-digit year setting was
selected in the OS. Some did roll over the dates to make sure the OS kept up.

MS-DOS and all versions of Microsoft Windows rely on either the BIOS or the hardware
for date information at start-up. Windows 9x can even correct an erroneous 1900 date to
2000, but if the BIOS is not updated or patched by the end of the year, Windows 9x will
think it is 1901 next year.

Systems running Windows NT ignore the BIOS altogether and get date information directly
from the hardware layer. If for some reason the real-time clock shows a year between 00
and 20 while the CMOS says the century is 19, NT will update the entry to 20.

That said, other OS components do rely on BIOS dates. Microsoft has released patches
for Windows 9x and NT that correct the components; check its site at http://www.microsoft.com/year2000 to see
whether you need them.

At Layers 4 and 5, McAfee 2000 ToolBox and Check 2000 PC showed limited scanning
ability. Check 2000 PC Deluxe and Norton 2000 did true scanning of applications and data.

Most Layer 4 tests compare the applications installed on the computer against a
database of the most recent readiness information provided by software publishers. In
Layer 5, the utilities mostly look for two-digit year fields and suggest updating them to
four digits.

Norton 2000 merited its Reviewer’s Choice designation by performing admirably in
Layers 4 and 5, especially under Windows 9x.

The amount of scanned date information from any actively used PC can be overwhelming.
Norton 2000 lassoed it into an easily understood format that mostly listed spreadsheets
and databases with two-digit year fields as problematic.

Although Norton officially supports only Windows 9x and NT, a handy component under
Windows 9x creates a bootable floppy that can test layers 1 and 2 under almost any OS on a
Intel-based or compatible system. Unfortunately, Norton cannot install its patch from the
floppy.

Like many Symantec products, Norton 2000 can use the LiveUpdate utility to download the
most current information from the Web. I wish it fully supported MS-DOS and Windows 3.x.
Norton 2000’s interface is somewhat confusing, and the user must scan the PC at every
launch before seeing results. The tool should make data from the previous scan available.

Government buyers can get Norton 2000 from the National Institutes of Health Electronic
Computer Store II contract for $20, which is $30 less than the open market price.

The $20 Y2K Test&Fix does support MS-DOS and Windows 3.x and is in use by the state
of Washington for year 2000 testing. Beyond layers 1 and 2, the SecureNet product
doesn’t do much. I consider its no-frills approach a negative, but a busy systems
administrator might consider it a plus.

CheckIt 98 Diagnostic Suite, priced at $130, performs year 2000 testing only as a
sideline. TouchStone needs to add components to its clear interface, which gives the user
little year 2000 information. Why can’t the interface say whether it checks that Feb.
29, 2000, rolls over correctly to March 1?

There’s hardly a difference between Check 2000 PC and Check 2000 PC Deluxe. Check
2000 PC, which works under Windows 3.x and loads from two floppies, is the most complete
tester for PCs running the older OS. In general, Check 2000 performed solidly, although
its test for Sept. 9, 1999, turned out to be bogus.

Doomsayers claim that date could cause catastrophe because, in hexadecimal terminology,
9999 stands for “end of file.” This might be a legitimate concern for older
mainframes, but PCs read Sept. 9, 1999, not as 9999 but as 090999.

The real-time clock has two characters in each time and date field.

Both Check 2000 packages did limited scanning of applications and data, but their
spreadsheetlike results were confusing. Plus, the suggestions were vague and not that
helpful.

Check 2000 PC faults any apps, especially those from Microsoft, that assume two-digit
years from 00 to 19 are preceded by 20.

When a user enters 02/02/02, the apps assume Feb. 2, 2002. I don’t think
that’s such a big deal.

Check 2000 PC also forces its windows to the front, obscuring a chunk of the screen.
You cannot get around this by setting it to scan, then working on something else.

The Deluxe package adds a separate application to scan data. Whereas Norton
intelligently chooses which files to scan, Check 2000 PC Deluxe might spend a long time
scanning every file on the hard drive.

The data scanner can be targeted to specific file extensions, but the results might be
too hefty to sort through or resolve. Also, I was uncertain where to begin fixing the
flagged files.

ProveIT 2000 has potential as a solid performer, but the interface needs help. To run
ProveIT, you must insert a floppy into your PC. Also, each of the four date tests requires
clicking a button instead of launching all at once.

McAfee 2000 ToolBox seems to have been pulled together from separate applications. It
has a virus scanner—not a bad idea.

But it performs diagnostics backwards: data, virus, applications and then hardware and
BIOS.

McAfee advertises updates to 2000 ToolBox on its Web site, but I could not get to the
patch without registering for a secret code. I also couldn’t get the live-update
feature to work.

McAfee 2000 ToolBox also installs WinGauge Lite, which tests system tuning, but an
obnoxious alert frequently warned that the hard drive was 9 percent fragmented even after
complete defragmentation. I finally disabled it.

If you need a serious tool for PC testing, Norton 2000 is the best choice. For older
PCs running MS-DOS or Windows 3.x that might need a third-party BIOS fix, go with Y2K
Test&Fix.