Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Reading the Microsoft TechNet article "Test results: Extra-large scenario (FAST Search Server 2010 for SharePoint)", I came across this graph.

enter image description here

Now, am I stupid (and if so, could someone please enlighten me), or is this graph just stupid? To me, it seems like it's created with MS Paint instead of reflecting real values - I can not make sense of the graph as it seems like any given value on the horizontal axis (Queries per second) can result in multiple values on the vertical axis (Latency).

EDIT: Thanks to all who contributed, the curves (not graphs) make more sense to me now. My conclusion to the original question is somewhere in the middle. I admit some stupidity on my own behalf, but still think the graphical representation of the data is quite stupid, as it fails to clearly communicate what it should.

share|improve this question
1  
Is it just its lack of visual appeal that you consider "stupid", or is there some other issue with the graph that leads you to that conclusion? –  Arturo Magidin Jun 20 '11 at 21:05
    
It is my lack of understanding how any given value along the horizontal axis (Queries per second) can result in more than one result along the vertical axis (Latency). –  Christian Nesmark Jun 20 '11 at 21:08
1  
@Christian: Could I then ask you to write that instead of using the ambiguous "stupid graph" descriptor? Perhaps something along the lines of "I cannot make sense of this graph: how can a single value of "queries per second" result in more than one value for the latency?" –  Arturo Magidin Jun 20 '11 at 21:11
    
@Arturo: Good point - I have added that now :) –  Christian Nesmark Jun 20 '11 at 21:18
    
@Arturo: But still, have you ever seen graphs like these? Can you make any sense of them, or can we maybe assume something along the line that whatever tool was used to generate the graphs has a bug in it? –  Christian Nesmark Jun 20 '11 at 21:23

2 Answers 2

up vote 2 down vote accepted

The only mathematics here is whether these curves represent functions, which they do not. They represent data taken as a function of time. these curves just show the behavior of the system with increasing time.

share|improve this answer
    
I agree to this, but why then use a graph to represent the data instead of individual sample points? Do these graphs tell you anything useful? –  Christian Nesmark Jun 20 '11 at 21:31
    
The way reason they chose a line instead of points is so that you can "follow" along and see the evolution of the system. Otherwise you would come to a point where there are two y-values for an x value and you would not know what that meant. Again, think of time evolution. If you want to know why the graphs curve backwards, from article," the farm was able to sustain about 15 queries per second (QPS) with less than 1 second average latency before being limited by the CPU resources on the search row. During incremental crawls and full crawls, this number dropped to 12.5 and 10 QPS respectively" –  Edison Jun 20 '11 at 21:42

Reading the article you mentioned, it seems to me that they did multiple runs of the same question, which is why you get multiple values for a given number of queries per second. What seems very unorthodox is that they polished that by drawing the lines which seems to reflect the timing of the measures (you get sort of a dynamic trajectory of the system).

share|improve this answer
    
Sounds reasonable. –  Christian Nesmark Jun 20 '11 at 21:34

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.