From: Rodgers Moore (rodgers@xxxxxxxxxxxxxx)
Date: Sat Apr 07 2001 - 18:23:51 GMT-3
Oops, ignore the side note, I typed quicker than I thought. :)) 250ms is abou
t
double the minimum delay of 125ms. However, the delay from Florida and Texas w
ill be
greater than 125ms as neither are on the equater. There are cases where the 25
0ms is
used, but not in this case.
To roughly calculate the delay from any spot on the earth. Take a globe, ruler
and
string. Find the scale of the globe. Let's assume 1000 miles = 1 inch. Exten
d the
ruler (yard stick) perpendicular to the equater from the spot directly below th
e
satelite. Then run the string from the 22 3/4 inch mark on the ruler to the sp
ot on
the earth. Now measure the length of string and multiply the inches by 1000 to
get
the miles. Take this number and divide by 186. The result is approximate to t
he
number of milliseconds it takes a signal to traverse that distance.
186? speed of light = 186000 miles/sec = 186 miles/msec If we divided by 186
000
we'd get seconds, not milliseconds.
Rodgers Moore
Rodgers Moore wrote:
> I guess you have to assume no error rate, but this is actually a pretty simpl
e
> problem. I assume the what they are saying is the MTU is 32K.
>
> 1.544Mbps / 8bit/Byte = 193000Bps
>
> Window size is 64K + (64K/MTU (32K) * 20Bpp) = 64040Bytes per window
>
> 193000Bps / 64040B/Window = 3.01 windows/sec or 3 ACKps
>
> Propogation delay is 250ms one way and an ACK is 250ms in the other direction
. So
> delay is 500ms per ACK or window.
>
> 3 ACKs per second * 500 ms = 1.5 seconds of delay per 1 second of raw
> transmission.
>
> 2.5 seconds total time to transmit 1.544Mb
>
> 1.544Mb / 2.5 s = 617.6Kbs throughput
>
> As a side note, in reality the delay from earth to satelite is about 250ms. T
he
> delay through the satelite is about 25ms, so earth to satelite to earth is 52
5ms
> so in truth the ACK delay is more like 1.05 seconds.
>
> Rodgers Moore
>
> Muthu wrote:
>
> > Hi Gurus,
> >
> > Found this on one of the CCIE prep sites...could someone help me solve this
?
> > and/or also direct me to the URL for such info...
> >
> > Question:
> > The TCP protocol does all its sequencing and window accounting in single by
te
> > units, although data may be transmitted in segments of upto 32KB. The maxim
um
> > window size provided for the protocol is 64KB. What is the approximate
> > theoritical maximum one way throughput of a single TCP connection over a
> > 1.544Mbps satellite connection from Dallas to Florida with a one way delay
of
> > 250 milisecs?
> >
> > Thanks in advance,
> >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:29:42 GMT-3