Information Technology Reference
In-Depth Information
4
3
2
1
0
2
4
6
8
10
12
Number of Servers
Concurrent Push
Concurrent Push w/AGSS
Concurrent Push w/AGSS & SSS
Staggered Push
Figure 12.10
Client buffer requirement versus system scale
push has a constant client buffer requirement that will not limit scalability. Note that although
client buffer requirement in concurrent push can be controlled to a constant by SSS (cf. Section
10.5), the system scalability is still limited as client processing overhead due to SSS increases
with system scale. It is particularly important to maintain a constant client buffer requirement
in practice as it would be very expensive (if not impossible) to upgrade every existing client
devices (e.g., set-top box) whenever the system is scaled up.
In Figure 12.11, we analyze the sensitivity of client buffer requirement to server clock jitter.
As the results indicate, the buffer requirement is relatively insensitive to clock jitter, even if
the jitter is increased to one second. Hence one can safely employ the existing software-based,
distributed clock-synchronization protocols in staggered push.
12.6.4 System Response Time
Figure 12.12 plots the system response time versus system scale. While the worst-case system
response time increases linearly with more servers, the average system response time remains
low (
2 seconds) for a utilization of 90%. This suggests that we can maintain a low system
response time simply by limiting the system to, say, 90% utilization through admission control.
In Figure 12.13, we study the sensitivity of system response time to server clock jitter.
As expected, the system response time increases for larger clock jitter values (cf. Theorem
12.1). However, given that server clock jitter can readily be controlled to within 100ms [1],
the average system response time is still only 0.735 seconds for an 8-servers system at 90%
utilization.
∼
Search WWH ::
Custom Search