Information Technology Reference
In-Depth Information
100
10
1
0.1
0
0.2
0.4
0.6
0.8
1
System Utilization
Reload from back-up
Baseline rebuild
Distributed rebuild
Mixed distributed-baseline rebuild
Figure 14.8
Rebuild time versus system utilization
scalable. For example, under a system utilization of
5, a mixed distributed baseline
rebuild can complete the rebuild process in 2.61 hours and 2.87 hours for a 16-server and
64-server system, respectively. Thus, if no new server failure occurs within this period, the
system can then resume normal operation and be ready to sustain a new server failure in the
future.
ρ
=
0
.
14.8 Summary
In this chapter we addressed the issue of rebuilding the data lost in a failed server to a spare
server so that the systemcan resume normal operation. Using amodel incorporating the transfer
capacity of the servers, the system utilization, and other system parameters, we showed that the
mixed distributed-baseline rebuild algorithm can achieve the maximum rebuild rate. Moreover,
the algorithm is completely automatic, and transparent to existing streaming sessions. The
numerical results showed that such a data rebuild process can be completed in a reasonably
short time (a few hours) and thus can enhance the reliability of the system.
Search WWH ::
Custom Search