(e) Exclude or block the source of the threads
(f) Restore the best-fit configuration of hardware
(g) Restore best-fit of configuration of system software
(h) Restore best-fit of configuration of application software
Moreover, making DCS fit for RT and safety-critical applications one should follow
the same procedures established in the GAFT framework with a little bit of extra
work:
(a) Analyze possible damages along with estimation of potential consequences for
network topology as well as elements,
(b) The speed of thread propagation in the network is a factor that delineates the
complexity of recovery procedures,
(c) Different segments of the network might have different importance for the
network as a whole, for instance a gateway host and a simple internal router
have different roles in the network.
Therefore, the consequences of damages resulting from threads might be different,
occasionally substantial and exponential dangerous if the problem it is not treated in
an effective manner. Restarting and segmental switching are solutions which are
1
2
4
5
8
11
10
6
7
3
9
Fig. 18.2 Connected computer topology (a fragment)
256
18 Distributed Systems: Maximizing Resilience