Tim,
Your message raises an interesting (but mostly unrelated) point;
>You need to add the memory at the bottleneck router, which is not
>necessarily the uplink router if the satellite link is not bottleneck.
>(But the adminstrator of the bottleneck router may have no connection
>with the adminstrator of the satellite link, so this is a problematic
>approach to solving this problem.)
It really hasn't been discussed here on the list, but one can claim
that even large-window extensions are a "problematic approach to this
(throughput) problem".
Look forward to the day (hopefully real soon) when window-scaling
is ubiquitous; As we remember, window-scaling does *not* mean
large-windows (it allows it to happen, but the buffer space needs
to be allocated on BOTH sides of a connection). Now, imagine:
o You want to do something on the Internet that will benefit
from large windows (say: transfer size > bandwidth*delay product)
o You are clever and *know* that large-windows will be beneficial
for you
o You make sure that you application (or system defaults) provide your
connection to a distant server with enough buffer space to advertise
the necessary large window
o You manage to establish a connection, and guess what... the server
only advertises a window size of 32KB (or something smaller than
you need)
Bummer, despite all your clean living, you're in Pokeyland! :o(
Also, if you are *not* clever enough to know that you should be using
large windows (maybe because it's not the default for your box, or
maybe you just don't know anything about the path), you're toast no
matter what the server has available :o(
So, can we *really* expect various servers on the Internet to change
their default buffer allocations to accommodate long-delay paths at
reasonable data rates? Do we even want them too? What's a "reasonable
data rate"?
Memory might be cheap, but so what? Does that give me a right to tell
anyone to increase their default buffer sizes by an order of magnitude
(or two)?
In addition, if we try to run with windows that are greater than the
bandwidth-delay product, we'll be continually knocking on congestion's
door and pay for with a performance hit. Seems like a fairly anti-social
thing for us to do regardless (people will just hate us).
Even if we control both endpoints to a connection (and maybe a large
chunk of the path between), there will be no single silver-bullet for
improving performance in this environment. They are all going to be
problematic.
Darn! :o)
Eric
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:29 EST