Par le aison je connaitre le stupidity de le mos je cintact visual ware direct. Par le raison cett organstation est tres concerne avec cette commentaire il repondre plus vite.Je envy le totale de cette comment en francais pour leur comment et gagne un comment plus vite par cette raison il en anglais. Pas que, toutle monsieur dit est stupid et ausi totalement fause:- le M ecrire poriginale par Phildur les response par le grand agence Visualware est suivant!:-
Thank you for the heads up.... Interesting... The statements being made are
not valid at best and ignorant at worst... clearly the person does not
understand TCP. I am a TCP engineer I spent 22 years writing embedded TCP
stacks. I am not entirely sure of the point being made by the author but the
mistake being made is that the he believes megabits per second is a speed.
It is not, mpbs is just a rate.
If you asked what time you should leave to catch a 10am flight and I replied
the road to the airport operates at 4000 cars per hour, would that help
you.... no.... what you need to know if how long will it take you to reach
the airport.
What matters is time, and to put that in context, as an example, it should
be noted that a 20ms 1Mbps and a 20ms 1Gbps connection operate at the same
speed. More importantly a 1Mbps connection can easily outperform a 100Gbps
connection.
Visualware tests report 2 numbers that are important no mater where you are
relative to the testing server. Download speed and Route speed. Route speed
defines the TCP equilibrium for the latency of the connection. The download
throughput speed must always be within 6%. Of this metric is the connection
is good.
Now the topic of TCP equilibrium is a lengthy one I could type several
pages. However suffice it to say that the first rule of an accurate measure
is to understand what is being measured and what it means. As an example for
thought, consider:
2 taxis are taking people from a conference to the airport 30 miles away.
The speed of the road is 60mph. Each taxi does 2 trips. One taxi is larger
and takes 8 passengers the other taxi takes only 4 passengers. The larger
taxi has a problem and has a top speed of 30mph, the 4 passenger taxi can do
60mph. The larger taxi therefore take 1 hour to get to the airport (i.e. a 2
hour round **bip**) the smaller takes 30 minutes (i.e. a 1 hour round **bip**)
After two trips the larger taxi has delivered 16 passengers in 4 hours for a
rate of 4 passengers per hour, the smaller taxi has delivered 8 passengers
in 2 hours for a rate of 4 passengers per hour. BOTH taxis report the same
rate, but the passengers in the large taxi took twice as long to get to the
airport and missed their flights, the smaller taxi was twice as fast.
What this example demonstrates is that the two cars do 60Mph is not the same
as one car doing 120mph. You cannot add 2x60 and report that as a
performance of 120.... Like it or not that is how TCP works, what matters is
trips per second not bits per second. The smaller taxi was twice as fast...
1Mbps can out perform 1Gbps or even 10Gbps...
If anyone is having a performance problem if they send me their Visualware
test result I will be happy to advise on the problem..
I hope this help and I appreciate the heads up Roger... I have typed this
quickly, please excuse any typos....
Thank you
Julian Palmer
Visualware Inc.
