Performance Modeling and Benchmarking of Event-Based ... - DVS
Performance Modeling and Benchmarking of Event-Based ... - DVS
Performance Modeling and Benchmarking of Event-Based ... - DVS
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
5.1. SPECJMS2007 - A STANDARD BENCHMARK 91<br />
180<br />
Transactional Non-Transactional<br />
No. Locations<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
5 15 25 35 45 55 65 75 85 95<br />
HQ<br />
SM<br />
SP<br />
DC<br />
100%<br />
90%<br />
80%<br />
70%<br />
60%<br />
50%<br />
40%<br />
30%<br />
20%<br />
10%<br />
0%<br />
P2P Pub/Sub P2P Pub/Sub<br />
Base<br />
% Msg. % Kbytes<br />
Figure 5.5: # Locations for Horiz. Topology<br />
Figure 5.6: Horiz. Topology Message Mix<br />
Figure 5.7: Proportions <strong>of</strong> the Interactions<br />
based on Msg. Throughput<br />
Figure 5.8: Proportions <strong>of</strong> the Interactions<br />
based on Msg. Traffic in KBytes<br />
λ 1 λ 2 λ 3 λ 4 λ 5 λ 6 λ 7<br />
1.53920154 2.13333333 6.00000000 3.37837837 11.54401154 11.38519924 9.23076923<br />
Table 5.4: Interaction Rates for the Horizontal Topology<br />
traffic per location (<strong>and</strong> therefore also per destination) remains constant. In the Figures 5.7 <strong>and</strong><br />
5.8 the relative weights <strong>of</strong> the interactions is illustrated. They are set based on a detailed business<br />
model <strong>of</strong> the supermarket supply chain which captures the interaction interdependencies. This<br />
model has several input parameters (e.g. total number <strong>of</strong> product types, size <strong>of</strong> supermarkets,<br />
average number <strong>of</strong> items sold per week) whose values are chosen in such a way that the following<br />
overall target messaging mix is achieved as close as possible:<br />
• 50% P2P messages <strong>and</strong> 50% pub/sub<br />
• 50% <strong>of</strong> P2P messages persistent, 50% non-persistent<br />
• 25% <strong>of</strong> pub/sub messages persistent, 75% non-persistent<br />
The goal is to put equal weight on P2P <strong>and</strong> pub/sub messaging. Within each group the<br />
target relative weights <strong>of</strong> persistent vs. non-persistent messaging have been set according to<br />
the relative usage <strong>of</strong> these messaging styles in real-life applications. The criteria for what is<br />
a typical MOM application were defined based on input provided by the various participating<br />
vendors in the SPECjms working group including IBM, Sun, Oracle, BEA, Sybase <strong>and</strong> Apache.<br />
A comprehensive survey was conducted considering real-life customer applications <strong>and</strong> analyzing<br />
their workloads.<br />
Table 5.5(a) shows the resulting message mix in the horizontal topology. Figure 5.6 presents<br />
the same data in graphical form. Figures 5.9 <strong>and</strong> 5.10 show how the number <strong>of</strong> messages <strong>of</strong> each