Ping Huang <pshuang @
sgihub .
corp .
sgi .
com> said:
>A packet collision rate on an Ethernet probably
>results in only slightly degraded performance for users; however, a 5%
>rate of failed connections (and there were several combinations of
>vendor products and the number of established virtual circuits in
>Table 3 in the article which had that loss rate or higher) would mean
>that about 5% of attempts to access a given home page will time out,
>and a much higher percentage of Web pages will be displayed with a
>missing inlined graphic.
>[...]
>Table 3 in the article which had that loss rate or higher) would mean
>that about 5% of attempts to access a given home page will time out,
>and a much higher percentage of Web pages will be displayed with a
>missing inlined graphic.
a) "3 out of 1000" is much less than 5%. You're off by a factor of 16.
b) this kind of load-related failure scenerio is where packet filtering
can really shine, especially when coupled with the ability to insert
the 'features' of proxies into the connection startup. (And then to
not have the proxies 'in the way'.)
c) NTSL's testing methods leave much to be desired. (Don't make me tell
you about the grot they spoon out in order for a PC to be 'NTSL certified'.)
d) I don't like how you got from dropped packets (or even closed sessions)
to missing inlined graphics. Don't bother explaining, I think I
grok your train of thought, I just think its a bit over-selective.
How do we know that the firewall(s?) that looked constipated were not, in
fact, sending back some flavor of ICMP message that resulted in the dropped
sessions?
Jim
Follow-Ups:
|
|
