This high integrated system has been build for a big
customer from the financial sector and is used to
synchronize 17 physically separated networks,
consisting of several hundred servers and thousands
of clients. Due to the fact that an extra housing
with up to 16 additional server modules can be
connected, the maximum stage of expansion can be 33
servers. The required space equals 6U in a standard
19" rack, the extension case adds another 3U to this,
if it is installed.
Front view
The LAN-CPU modules and the SYS-CPU module each
represent a full featured NTP server and will be
synchronized by the secure hybrid system, which
permanently compares the two time signals. If the
difference between them exceeds a configurable limit,
a so-called time limit error will be issued and the
clock system stops issuing the time to its
LAN-CPU/SYS-CPU modules. This provides a very good
protection against the manipulation of the GPS and/or
DCF77 radio signals. An operator has to reactivate
the time synchronization when this happens.
If the primary receiver (in this project the
customer wanted to use the DCF77/PZF receiver as the
primary time source) fails, the secondary receiver
takes over. In case both receivers fail, the rubidium
oscillator is used to ensure time stability for the
time which is needed to restore operation of the two
radio clocks. In "normal mode" (i.e. when both
receivers are synchronized properly), the rubidium is
disciplined by the GPS receiver.
Rear view
The SYS-CPU module is connected to each of the
LAN-CPU modules over a limited serial multiplexer and
can only read status informations and log entries.
This ensures that the networks are separated from
each other. The configuration of each LAN-CPU module
therefore has to be done on the module itself, first
by connecting a terminal and using the console setup,
after setting basic network parameters the standard
LANTIME configuration options (e.g. SSH, HTTPS or
SNMP) can be used (Demo).
|
