Outliers in residuals plots for Anova I've created an Anova model comparing my algorithm and another algorithm. My experiments are 3 clients accessing 3 servers, with 10 Experiments each Algorithm.
I didn't understand how to explain the distribution of my residuals.While my Normal Q-Q Plot looks pretty well distributed closely to the line. The Residuals vs Fitted shows a huge number of outliers and the residuals plot shows a strange result.


 A: this would suggest that four of your readings/values are exceptional and need attention. You can either adjust them for their "outliers" or introduce four dummy variables each a 0/1 to moderate their effect using regression analysis. These 4 values ( as observed) will defeat your ability to correctly estimate the impact/significance of your design variables. These four values should be investigated as Bacon pointed out ....
To do science is to search for repeated  patterns.
To detect anomalies is to identify values that do not follow repeated patterns. 
For whoever knows the ways of Nature will more easily notice her deviations
and, on the other hand, whoever knows her deviations will more accurately
describe her ways.                                                              
One learns the rules by observing when the current rules fail.      
These four observations are being effected by variables outside of your "knowledge" . While it is good practice to adjust these values it is even better practice to find out what actually "caused" these exceptional values.            
