Are you sure BIC-TCP is in network layer? I assume that they are not messing with the Network Layer (IP) and BIC-TCP replaces TCP in the transport layer.
However, I did not read the paper so I could be wrong. Especially in assuming that BIC-TCP runs over IP
The problem with Cable + VOIP is QoS. When your cable goes out, no phone. This includes power outages - and even if YOU have UPS or whatever, the likelihood is that your cable headend doesn't have anything but a few minutes of battery.
The Headend/Hub will have power to last for days, using very large diesel generators. The network (amps etc) will have a much smaller power supply. Many (well at least where I used to work) have power supplies that last between 3-6 hours, and if those start to go we would roll a truck and setup a small gas generator to power the amp.
So the case may not be grim, not saything that there are not headends who have old crappy headends that might only last a couple hours, depends on where you live.
From what I've heard [anecdotally] computers like the earth simulator go vastly under utilized for the most part.
From my experience that is mostly untrue, yet widely publicized. Yes, if you look at utilization as the (used-proc*sec)/(totaltime*numprocs) the number can be relatively low (~60-70%). However, that includes system time, rebooting the machine, weekends, holidays, etc. Further, when it comes down to it the researchers need to have a reasonable turnaround time during the day for their development runs (when the utilization is much higher than 60%). Further, since these machines generally run jobs of different sizes from many differnt users there is an upper bound on utilization
Slashdot Mirror
Are you sure BIC-TCP is in network layer? I assume that they are not messing with the Network Layer (IP) and BIC-TCP replaces TCP in the transport layer. However, I did not read the paper so I could be wrong. Especially in assuming that BIC-TCP runs over IP
The problem with Cable + VOIP is QoS. When your cable goes out, no phone. This includes power outages - and even if YOU have UPS or whatever, the likelihood is that your cable headend doesn't have anything but a few minutes of battery. The Headend/Hub will have power to last for days, using very large diesel generators. The network (amps etc) will have a much smaller power supply. Many (well at least where I used to work) have power supplies that last between 3-6 hours, and if those start to go we would roll a truck and setup a small gas generator to power the amp. So the case may not be grim, not saything that there are not headends who have old crappy headends that might only last a couple hours, depends on where you live.
Not to be picky, but why boot to linux? Why not just open a terminal from osx and tar it?
From what I've heard [anecdotally] computers like the earth simulator go vastly under utilized for the most part.
From my experience that is mostly untrue, yet widely publicized. Yes, if you look at utilization as the (used-proc*sec)/(totaltime*numprocs) the number can be relatively low (~60-70%). However, that includes system time, rebooting the machine, weekends, holidays, etc. Further, when it comes down to it the researchers need to have a reasonable turnaround time during the day for their development runs (when the utilization is much higher than 60%). Further, since these machines generally run jobs of different sizes from many differnt users there is an upper bound on utilization
ummm...no....I think it is called sarcasm...