The AI answer we saw
Both Google Cloud Metadata Server and Azure IMDS operate at the hypervisor layer via the 169.254.169.254 link-local address, resulting in sub-millisecond to low single-digit millisecond latency for most requests under normal conditions. Neither service traverses the external network, so latency is largely independent of regional network conditions. Google Cloud's metadata server also supports the 'metadata.google.internal' DNS alias, which may add minor DNS resolution overhead compared to direct IP access. Azure IMDS latency is similarly low, with Microsoft recommending caching of tokens and metadata to minimize repeated calls. Availability for both services is tied to the health of the underlying VM host; both providers design these services for continuous availability during normal VM operation. For latency-sensitive workloads, caching retrieved metadata and credentials locally within the application is a best practice on both platforms.
Captured during this Readable report run. Answers can vary by AI tool and date.