Differentiating data collection for cloud environment monitoring

In a growing number of information processing applications, data takes the form of continuous data streams rather than traditional stored databases. Monitoring systems that seek to provide monitoring services in cloud environment must be prepared to deal gracefully with huge data collection without compromising system performance. In this paper, we show that by using a concept of urgent data, system can shorten the response time for most `urgent´ queries while guarantee lower bandwidth consumption. We argue that monitoring data can be treated differently. Some data capture critical system events, the arrival of these data will significantly influence the monitoring reaction speed, we call them urgent data. High speed urgent data collection would help system to act in real time when facing fatal error. On the other hand, slowing down the collection speed of others may render more bandwidth. Then several urgent data collection strategies that focus on reducing the urgent data volume are also proposed and evaluated to guarantee the efficiency.