DART Version 4.2
----------------

Welcome to DART . . . . .

DART stands for Data Access Reliability Test and it is
designed to serve two purposes.  The first is to provide
a means of benchmarking all components in a network, from
the user PC and LAN adapter to the server PC, operating
system, disk cache, LAN adapter, disk controller and
drive.

The second function is to test the overall data integrity
of the network.  Most benchmarking systems rely purely on
raw data throughput, not bothering to check the data.
DART works by writing a user specified number of random
data records.  It then works back through this data file
checking that what was written is read back correctly.

Although this may seem unnecessary, we have encountered
several systems that have failed this test - mainly
through Netware driver/OS bugs and intermittant hardware.

DART is a multiuser testing system, supporting up to ten
simultaneous stations.  DART has been tested on a wide
variety of networks including Netware 2.x, 3.1x, 4.x,
Netware Lite and LANtastic.  DART will also run on a
standalone PC, useful for finding intermittant disk or
controller problems.

To facilitate multi-user operation one PC must be
allocated as the MASTER and all others run as SLAVES.  To
start the master PC, run MASTER.  To start any slaves,
just run SLAVE.

DART provides a performance figure (Records Per Minute)
which reflects the overall performance of the system.  As
the number of different components in a system can vary
widely, it would be impractical to publish a set of
sample benchmarks.  The best we can suggest is to use
DART regularly and track the results so that you can
detect when a problem starts to arise.

DART can be set to use from 1 to 3000 records per test
for each workstation.  Each record takes up about 1KB of
disk space so running ten workstations at 1000 records
each can chew up the disk space (1000 x 10 x 1000 =
10MB).

In general, we use the standard setting of 1000 records
for initial testing.  However, for complete testing of
the system you should use at least enough records to
completely use up all server/cache memory.  This will
ensure a thorough test of the fixed disk and controller.

So, for a server with 6MB free running four workstations
you should set the test to about 3000 records (3000 bytes
x 4 stations x 1000 records = 12MB).

Disclaimer
----------

DART has been designed for a certain use and has been
tested for compatibility on a number of hardware types.
We cannot, however, guarantee that it will work with any
particular configuration or guarantee it's performance,
even if used for it's intended purpose.  We will in no
way be responsible for any damage that may result from
the use of this software.

The software is supplied on an as-is basis and you use it
at your own risk.  We do request that you notify us (at
the address below) with any problems you encounter but we
are under no obligation to provide a remedy for any
reported problems.

License Terms
-------------

DART is supplied as shareware and you may use it for any
purpose you feel suitable, on as many PCs as you like.
You may re-distribute DART in it's original ZIP file and
only if this file accompanies it.  In any case, you may
not modify the program or this documentation in any way
other than for your personal use.

If you feel that DART is useful and you continue it's use
we would appreciate a small payment to assist with future
development.  We suggest $5.00 for a single user, $50.00
for a site.  In return for this payment you will receive
an update to the latest version.

Source code for DART will be supplied with any donation
over $50.00 and a sample set of other utilities by our
company will be supplied in return for any payment over
$25.00.

Postal address:          LAN Technologies
                         PO Box 27-424
                         Mt Roskill
                         NEW ZEALAND

Compuserve user ID:      100033,2241


