61

4.6

Performance Analysis

4

x 10

6

1400

5

ssl_session

ssl_bf

1200

RR

4

1000

3

ssl_session

ssl_bf

800

Throughput

2

RR

600

1

Average Request Latency (ms)

400

0

26

24

22

20

18

26

24

22

20

18

k

k

(a) Latency

(b)Throughput

Fig. 4.5. Average Latency and Throughput of a 16 node Application Server

The latency and throughput results of the three models (RR, ssl with session and

ssl with bf) in a 16 node application server are plotted in Figures 4.5. The results are

measured as k value decreases from 26 to 18, where k is the parameter of the Pareto

distribution for the incoming request interval, P (x) =  k

/x

 +1

. When k decreases, the

request interval becomes shorter, and consequently the load on the servers increases. The

range of k values is chosen such that the average utilization of the servers lies between

50% to 90%. Beyond this point, since all of the servers overloaded, a load balancing

mechanism among the servers is unnecessary.