In an inline configuration, all traffic can be halted if Suricata is not available or if it becomes overloaded with requests. How could you reduce the likelihood that a Suricata queue backup or failure would halt all network traffic until corrected?