@akornich another possible scenario is when Google hosting infrastructure having its bottleneck hit (that does happen occasionally) and we keep continuing hitting it with many requests from potentially thousands of mobile clients without giving it any opportunity to recover it makes things worse.
@akornich if i remember correctly it is critical to Google's load balancers.
@akornich An SDK instance should not send more than N payloads per minute as specified in local config.
Also, every HTTP reply from the server has rate limiting parameters within HTTP headers (on per time window basis).
@akornich At some point - absolutely have to. These are preventing/minimizing customers overcharges.
We need to throttle http transactions.