A next seminar  of Department   No. 2 on topic  ‘’Model for Detection of performance anomalies in cloud infrastructure in  real-time’’  was  held  at  the Institute of Information Technology.

A report was delivered by leading research fellow of institute , PhD Farqana Abdullayeva. In order to ensure the safety of cloud infrastructure detection of performance anomalies which can occur during the fulfillment of hardware, systems and software of this infrastructure assumes a great deal of importance.

F.Abdullayeva noted that Information security attacks in cloud environments creates anomalous behavior in memory, processor resources of the system,  and stressed the importance  of assessment of resource consumption state to detect anomalous behavior in real-time in the cloud system: " The implementation of this assessment  by following server  criteria  described with time series is advisable".

“Based on classifier ensemble idea, model is constructed on the basis of algorithms which is included to known classifier classes as Bayes classifier and decision trees’’- she noted.

Assessment of the effectiveness of the proposed model is carried out over an open cloud databases - " Google cluster trace " and "Yahoo! S5" and detection accuracy  is assessed on the basis of metrics, she  stressed

In the end, the reporter noted that, high results was achieved  during experimental verification of this model, anomaly detection accuracy in the course of testing of this model on existing databases is 99 percent and deviations forms  0.70 percent of the total data.

In the end, there was an exchange of views on the subject, many questions were answered.