this was pretty much how things worked in zabbix 1.0, and it did not scale in any way.
i doubt anything like this could appear - you could use trigger severities and other properties to classify reaction

That is why I said that the current global configuration should be a default for all triggers not specifically configured otherwise.

A little detail for why I rased this tiket (maybe I can configure this in 2.2.3 but did not figured it out yet):

I have the following requirements:

• Same trigger must send email notifications all the time by email to two teams (24x7). The same trigger must send SMS notification to a third team (work shifts). If the trigger is not acknowledged a forth team must be notified via email and SMS (escalation) (24x7).
• All other triggers must send only email (24x7) to first to teams all the time.
• All triggers have same severity

If the severity would be dynamic based on some criteria (frequency, ack delay etc.) then I see how a global alert configuration could be configured based on the severity but I can't figure out, with current features, how I can acomplish the above without scripting custom alerts.

And we are talking about a few hundred triggers from a few thousand that need the above behaviour.

Real world example:
Disk space monitoring: first trigger at 20% free space notifies operators and app admin, if not ack and not solved, before reaching second trigger at 10%, unix admin must be notified (sms). I still not resolved, incident management team should be notified (email + sms).

How do I configure this. Remember, is the same thresold (20%).