Information Technology Reference
In-Depth Information
700
DMPSS
Ferry−Pred
Ferry−Rnd
DMPSS+LB
Eferry
600
500
400
300
200
100
0
1
2
3
4
5
6
7
8
9
10
RP node ranking (by total matching time)
Fig. 2
The total matching time on each RP
Ta b l e 2
The number of received events on each RP
RP 1
RP 2
RP 3
RP 4
RP 5
RP 6
RP 7
RP 8
RP 9
RP 10
DMPSS
51331
51331
66655
81760
102400 102400 102400
Ferry-Pred
102400 102400 102400 102400 102400 102400 102400
Ferry-Rnd
102400 102400 102400 102400 102400 102400 102400
Eferry
51412
51412
66616
81909
79933
77401
87400
144019 72209
41574
DMPSS+LB 51413
51413
81839
66503
66503
102400 102400 102400
Ta b l e 3
The average matching time for an event on each RP(ms)
RP 1
RP 2
RP 3
RP 4
RP 5
RP 6
RP 7
RP 8
RP 9
RP 10
DMPSS 5.4916 4.7900 9.4579 3.5720 1.3814 1.1952 1.2437
Ferry-Pred 6.1936 5.9771 2.7091 1.7438 0.7186 0.1600 0.0828
Ferry-Rnd 2.7571 2.7232 2.5300 2.5285 2.4634 2.5908 2.4767
Eferry 4.6944 4.1609 6.5946 2.3502 0.9865 0.1004 0.0911 0.0900 0.1034 0.1450
DMPSS+LB 4.5712 3.8518 2.7472 4.0108 3.8806 1.0270 1.0218 0.9815
As mentioned above, the numbers of stored subscriptions can't evaluate the over-
head for matching comprehensively. So we calculate the total matching time that is
consumed on each RP during the event publication process, and the result is shown in
Fig. 2. Similar to results of Fig. 1, DMPSS realizes better even overhead for match-
ing than Ferry-Pred and Eferry, and Ferry-Rnd has the best matching load perfor-
mance. We record the numbers of received events and the average matching time
for an event on RPs, and list them in Table 2 and Table 3. As shown in the tables, in
DMPSS, RP 5, RP 6, and RP 7 store less than 400 subscriptions, and they receive
Search WWH ::
Custom Search