A virtual machine was migrated to another host, briefly pausing its hardware and accelerated NIC for about a minute. That interruption triggered a cascading failure in a replicated database setup and has been extremely difficult to recover from. Because the service was degraded rather than completely unavailable, I opened a Severity B support case under a Standard support plan. The stated initial-response target is less than four hours.
It has now been nearly 25 hours. The only response has been an automated AI-generated message containing generic troubleshooting suggestions that did not help. I selected 24/7 availability and requested a phone call, but nobody has contacted me or provided a meaningful update.
Does an AI-generated response technically satisfy the initial-response SLA, or should I expect a response from a human support engineer? We spend roughly $40,000–$70,000 per month on cloud services, and this experience has seriously damaged our confidence in the provider.
3 Answers
I recently had a Severity B production case with the same stated response target, and it took roughly six days to receive anything. The eventual advice was essentially to keep retrying until the service worked. Based on that experience, I would not assume that an automated message means a useful engineer response has actually happened, even if the provider counts it as the initial response internally.
There may also be a design issue separate from the support complaint. A one-minute interruption during host maintenance should not normally cause a replicated database environment to enter a prolonged cascading failure. It is worth reviewing the workload’s resiliency, failover behavior, and recovery procedures. Changing cloud providers may improve the support relationship, but it will not automatically eliminate the same availability risk.
If you need more reliable escalation, the Standard support tier may not be enough for that level of cloud spend. A higher paid support plan can provide 24/7 escalation channels, pooled technical contacts, and—at the top tier—dedicated contacts and proactive assistance. That still does not excuse a missed response target, but it may give you better ways to get a stalled case escalated.
We have decided to move to another provider where our sister company already has a strong relationship with a reseller that can escalate cases. We used to have an account contact here, but that person left and was never replaced.
That matches what we are seeing. A year ago we would at least get an overnight response; now cases can sit for days with only generic updates.