Current cloud systems do not consider the request volume according to the time, and they just maintain services through fixed number of servers. For preventing the system is being a security failure by internal and external threat, flexible reaction is needed.
To enhance the system security, we suggest a new approach, an adaptive intrusion tolerant system (ITS) based on historical data. This scheme utilizes historical data according to the time, predicts future condition, and adaptively adjusts the size of active VMs to improve the system performance and security level.
Proposed system showed 8.46 percent delayed response from whole requests, only 1.2 times higher response time than designated threshold, and 22 second action ahead in DoS attack condition. We expect this approach could be applied to various cloud system, and it will show higher performance and security level.