PICS is a rather large project, which is being performed by a surprisingly
small team. See what InfoWorld had to say about it in their 1995 Top 100
Client/Server Projects article.
The software consists of a large set of applications, DLLs and databases
which are distributed over a number of Windows NT machines. In general,
the all of the NT machines which are running critical system software work
in primary/backup pairs, so that the backup machine can take over the primary's
job in the event of a machine failure. Non-critical machines may run
as primary only, or with the support of a backup, but their absence will
not prevent the remainder of the system from running.
Realtime plant data is gathered by instruments throughout the plant,
then collected, converted and alarmed by our front end machines, which
are connected to the PICS via a separate network. This data is delivered
to a control node on the PICS network and redistributed to all of the PICS
nodes from there.
End-users may access the PICS data through bridge nodes. They will
receive some common plant data (like time, operating mode, etc.) and any
data they have requested. This keeps the overall traffic to the end-users
down to a tolerable level.
Currently, the PICS software consists of more than 40 separate applications,
and a dozen or so DLLs. The point definition databases are created and
maintained in Microsoft Access and then imported into the PICS database
offline. Eventually, we plan to add the capability to update the PICS
databases directly from Access while the system in online.
The PICS installation we are currently working on consists of a four
private networks, two for data collection and two for distribution, processing,
and display. These networks are also attached to the corporate WAN (via
custom bridging/proxy software), making the data available to end-user
desktops. Click here to
see a simplified (believe it or not!) diagram of the network configuration.