The current zabbix[queue] item is for the whole system and we trigger alerts on this, but we can't see if we have problems with one proxy, so a bad proxy sets off alerts (which also block unreachable alerts system-wide for us since we use that as a dependency) - much better to have a queue item per proxy (plus the hosts on the zabbix server itself separately) so we can track, graph, and alert on this. Otherwise when we have 10 proxies this is pretty useless, but we work globally so we often have latency and unreachable issues, so we need a queue to tell us when things are slow and inhibit unreachable alerts.