>> I'm curious, how does the ISP using a cache server save bandwidt? I can see it
>> if the requests are coming from their clients to web pages outside, but how
>> does it help when web page requests come from outside users to web pages that
>> are inside? The outside bandwidth is still used.
>
>Bandwidth inside their controlled network is used, but no additional
>traffic from the ISP through their core network to others' networks are
>used. Example: 20 customers in the same network request a page in
>*.yahoo.com. The ISP caches it after collecting it the first time. The
>next 19 customers receive a cached copy. Yes, bandwidth inside the
>network was used, but it was THEIR OWN network (read: cheap[er]). Those
>other 19 object requests did not have to cross transit networks.
What you stated above is exactly what I stated. You didn't answer the real
question, i.e., requests coming in from the outside to web servers inside
their network. External bandwidth is still used. All they end up doing with
their caching is pissing off their clients because updated pages are not making
it out to those requesting the pages.
>This is really getting off-topic and beyond scope. Please refer to
>those books I mentioned. You need a better understand of how the
>Internet "works" with respect to routing, and preferably, DNS. Google
>for split-horizon DNS, rather than wasting your time in this thread.
See a followup posting where I did Google split-horizon DNS, before you
suggested it. I do know how DNS lookups work and they have nothing to
do with actual packet routing. One can place the IP address directly into
the URL address line of the browser, totally skipping DNS lookup. Now we are
down to only how the HTTP packets get routed.
DNS domain name lookups and routing of packets are not related.
>> I do not disagree with ISPs trying to save money, but when a cahe proxy
>> works as badly as the one they are using, it has to hurt their customers.
>> If I were yuou, I'd be talking to their technical people right away to get
>> that problem fixed, because it is a problem.
>
>Not for the ISP, and not for their customers, in most cases. This is
>not a normal case. This is a situation where someone wanted to force an
>update, but couldn't (initially) get the behavior they expected. Yes,
>it was an inconvenience for Rodolfo, but that setting saves the ISP a
>[probably] healthy sum in connectivity costs (in-between carriers).
In the case of Rodolfo, his pages are INSIDE the ISP, which mean all outside
requests for his pages still use external bandwidth, so caching pages that
reside internal to that ISP save absolutely nothing in bandwidth. Instead,
grief is caused to their customer's web pages.
MB
--
e-mail: vidiot@xxxxxxxxxx /~\ The ASCII
\ / Ribbon Campaign
[So it's true, scythe matters. Willow 5/12/03] X Against
Visit - URL: http://vidiot.com/ / \ HTML Email
--
redhat-list mailing list
unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list
