and ave is the average inter arrival time of the incoming requests. We measured perfor
mance when the session period increases from ave to 100 ave. In Figure 4.8, we can
see that the latency and throughput of both the ssl with session and ssl with bf models
suffer from short session periods. At higher server load (k=12), ssl with bf has almost
similar performance as that of ssl with session at the ave point. This is because when
the session period is short, many requests from the same client need a re authorization
process. With high server load, the re authorization process saturates the servers. In
this situation, the load balancing between the nodes is not necessary. This explains why
the results show that a highly loaded server suffers more than a lightly loaded server
with a short session period.
In Figure 4.8 (a), the latencies of both models plunge at higher load (k=12) when
the session period changes from ave to 2 ave, and at the 2 ave point, the latency of
ssl with bf is 90% of the ssl with session latency. The ssl with bf latency drops until the
period is the 4 ave and slightly decreases until the 8 ave point. At this point, the
latency of ssl with bf is about 45% of the ssl with session latency.
With lower load (k=14), the performance result of the two models are much better
than those of k=12 and the latency drops until the 4 ave point. The throughput results
of ssl with session and ssl with bf in Figure 4.8 (b) show similar tendency as the latency
results. The ssl with bf model yields better throughput than the ssl with session model
over the observation window. The results in Figure 4.8 indicate that the session period
should be at least 4 ave to reap the benefits of session reuse.
Next, we plot the effect of the average file size on the two models in a 32 node
application server in Figure 4.9. The average latency of both models increases and