This article attempts to efficiently cope with the requirement for the high
-level intelligence network (IN), and proposes the realization of a service
control node by a large-scale distributed configuration composed of a larg
e number of processors. As the basic distributed configuration, the service
control function and the service data function are realized by separate pr
ocessor modules. Then, the integrated configuration with function and load
distribution is considered, where the load for those functions is handled b
y multiple processor modules. As to the realization of the distributed conf
iguration, the load imbalance among the processor modules, as well as the n
umber of dynamic steps in the database processing, are evaluated. In other
words, a comprehensive comparison is applied from the viewpoints of cost an
d performance. The result of the investigation is reflected on the distribu
ted memory database configuration, which is proposed as a configuration sui
ted to the high-level IN. As to unbalance among the processor modules, a me
thod is proposed in which the access rates of the customers are approximate
d by their order and an analytical evaluation is applied. As to the recover
y of the database, the recovery levels are defined according to the data cl
ass, and an efficient method of recovery is proposed by combining the acqui
sition of the checkpoint database with the interval of several minutes and
the small-capacity update log data. (C) 1999 Scripta Technica.