<br><font size=2 face="sans-serif">Hello!</font>
<br>
<br><font size=2 face="sans-serif">A small suggestion for an improvement
to zttest: some sort of histogram to show a broader range of the
results that are being returned. For example, on a test machine I
ran each of the following items in separate infinite loops at the same
time:</font>
<br>
<br><font size=2 face="sans-serif">ssh-keygen -b 8192 -t rsa -f /test.key</font>
<br><font size=2 face="sans-serif">dd if=/dev/zero of=/test.file bs=1024k
count=5000</font>
<br>
<br><font size=2 face="sans-serif">while in a third console I ran zttest.</font>
<br>
<br><font size=2 face="sans-serif">I did this twice each over several hours.
My results were encouraging: Best case was 100% (and from watching
the output from time to time there were lots of those), average was 99.90%
for one and 99.98% for the other, but the worst-case was troubling: 83%
for one, and 68% for the other!</font>
<br>
<br><font size=2 face="sans-serif">Of course, over several hours, there
were tens of thousands of results, and even a single bad result will throw
off the worst-case result. Hence, the request for some sort of histogram:
something that would show how *many* results were way off, and by
how far. Something that would show the nature of the bell curve I
would expect to get.</font>
<br>
<br><font size=2 face="sans-serif">Of course, I could probably parse the
raw output of the zttest command with something to plot this. However,
my unix-fu is not good enough to do that. Does anyone have a suggestion?
Or would this be valuable to have in the zttest command internally?</font>
<br>
<br><font size=2 face="sans-serif">Tim Massey</font>
<br>