for your example, one day would be about 3% of downtime. My understanding of their sla, for the services ive checked with an sla, a 3% downtime is a 25% credit

for your example, one day would be about 3% of downtime. My understanding of their sla, for the services ive checked with an sla, a 3% downtime is a 25% credit for the month's total, or $2500, assuming its all sla spend. In this outage's case you might be able to argue for a 10% credit on affected services for the month, figuring 3.5 hours down is 99.6% uptime. but i still agree, it cost us way more in developer time and anxiety than our infra costs, and could have been even worse revenue impacting if we had gcp in that flow