Technology considerations on large zabbix environment

[alaahil]

Hello All,

We are building a large zabbix installation with over 2000 machines spread over several networks. We have decided to put a proxy over each network. For the zabbix server and database we would like to get some feedback whether the approach we are following makes sense or not.

1- For the server we are planning to use the HA feature that is natively available in zabbix. However, we were advised to achieve high availability through virtualization, has anyone any experience in that manner? is it recommended?
2- For the the database we will do galera clustering with maria db. again there were suggestions to use master slave instead, but knowing that zabbix would be write intensive wouldnt galera be more efficient?
all advice are welcome

[cyber]

1. Not really sure, what you mean with "virtualization" but... why bother... out-of-the-box HA works just fine. Short switchovers etc.
2. Being write intensive... it really does not matter, it will still write to one and replicate to others. It sure will be HA and resilient and what not... but being faster? Unfortunately, I cannot even read their docs without submitting a request for access... Even that as starting point will drive me away from their solution..
I would go with PG and timescale... :P There's multiple additions to make PG replicate and switch over automatically, so you still get your HA... Patroni, pg_auto_failover (I have this in use).