Today, call control servers use COTS (commercial off-the-shelf) components such as Advanced Telecom Computing Architecture (ATCA) and Carrier Grade Linux. However, the adoption of COTS components causes an increase in the maximum response time of high-priority write system call during synchronous state transitions and recording whenever IO operation is congested by normal-priority IO requests. Call control is a real-time operation; the delay of state transition makes service quality worse and sometimes causes fail-over of service. The delays may occur due to failure of software and hardware. However, even if there are no failures, the response time of high-priority write system call sporadic delays due to insufficient consideration about the IO prioritization by the OS. We present reduction of bottlenecks of the file system operation and IO scheduler mechanism in the kernel to reduce the maximum response time of high-priority write system call. The evaluation demonstrates that the presented approach shortens the maximum response time of high-priority write system call with sufficient reliability.