Achieve high availability about point-single failures in OpenStack | IEEE Conference Publication | IEEE Xplore