voiprodrigo
commented
3 weeks ago This is a big concern for me as well. Currently in the process of evaluating a Datadog PoC, and the lack of ability to fallback to an alternate site/proxy is an architectural issue. With the Telegraf agent for example you can specify alternative urls for metrics delivery, and it will use them in round-robin and try the next one if one fails. The memory buffer can only help so much, and creating a traditional proxy ha setup is unnecessary work if the client supports at least dual-homing,
We have been using proxy between our DD agents and DD servers. Recently had problems with the proxy and all monitoring went really bad (all reds ...). It became obvious for me that the proxy becomes a bottle neck for alerting and monitoring once it is down. The solution would be to have several proxies to be specified in DD agent configuration.
Please add a feature to be able to specify several proxy servers in DD agent configuration.