Too many IPs assigned to an interface?

[Marc G. Fournier]

Since talking about ng_fec, and the cisco switch, I started to play with 
it a bit, and one of the things I've finally setup is snmp/mrtg, so that I 
can monitor bw activity ...

one thing that I've noticed is that two of my machines are doing alot of 
bandwidth, while the other two are doing significantly less ...

The thing is, the ones that are doing significantly less are the ones that 
have the most IPs assigned to their interfaces ...

based on 5 minute averages:

neptune -  68kb/s In, 119kb/s Out, 92 IPs assigned, Dual Xeon
mars    - 289kb/s In, 320kb/s Out, 35 IPs assigned, Dual PIII

vmstat 5 on neptune:

102 3 0 1722316 206436  258   0   1   0 465   0   4  49  511 3885 2398  3 86 12
102 3 0 1681208 205624   74   0   0   0  63   0   1   0  305 3293 1233  2 57 41
96 3 0 1702012 189492   69   0   0   0 845   0   6   6  342 3606 2066  6 53 41
91 3 0 1699380 151064   85   0   0   0 2072   0  12  12  418 2752 3239  9 23 69
90 3 0 1681276 148584   53   0   0   0 463   0   1   3  325 2554 2266  6 23 72

vmstat 5 on mars:

11 5 0 4071268 211624 2329   1   2   1 1348 486   0   0  710  378 1049  6 24 70
14 5 0 4059324 198648 597648   0   0   0 920   0  18 157  933 7267 12086  4 56 40
15 5 0 4070128 189200 652140   1   0   0 853   0   4 122  931 6188 9166  5 52 44
16 5 0 4056332 211964 693722   0   2   0 1690 1558   1 167 1276 5614 4517  9 49 42
16 5 0 4012580 208272 722681   0   0   0 1133   0   3 137  909 3839 5456  6 48 46

the other one that seems 'low' for traffic is a Dual Athlon (85 IPs) ... 
the other that is high for traffic is another Dual PIII (21) ...

So, is network performance that greatly affected by # of IPs assigned to 
the interface itself?  Or is there maybe another factor involved?

----
Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
Email: scrappy at hub.org           Yahoo!: yscrappy              ICQ: 7615664