Telcos and cloud providers that require a high degree of uptime and reliability aim to have less than 1% of incidents caused by code or configuration changes. In reality this is closer to 10%, but the goal makes clear the kinds of systems and infrastructure required.