Hello
I'm Joel Urtubia from Chile, I'm works on a telecom company on Chile, now we have monitoring 100K of devices with elasticseach with 8K values per second, but with this number the zabbix server was full, about the last we need to separate the load in two zabbix servers , ....we was talked with a collegue and read a lot information , and think that have tree ways to solved it:
1- use a zabbix proxy.
2- use an elasticsearch database independent for each zabbix server.
3- use an elasticsearch database but each zabbix server inserted on diferent index name
for solution 1, zabbix proxy use a local database for store the results , and is too slow for growing in devices, we was tested it.
for solution 2, this should be work but the main problems is that all data was separatly , is not scale for views and is difficult development a view for end customer.
for solution 3 , this should be the best solution, but I see that each zabbix only insert in the same index name, and it could be a problem... is posible that open a feature request??, for :
each zabbix server insert with a unique ID or name and will create a index, like this:
zabbixServer1 :
Index name elastic:
zabbixServer1-uint-date
zabbixServer1-dbl-date
zabbixServer1-str-date
zabbixServer1-log-date
zabbixServer1-text-date
zabbixServer2 :
Index name elastic:
zabbixServer2-uint-date
zabbixServer2-dbl-date
zabbixServer2-str-date
zabbixServer2-log-date
zabbixServer2-text-date
it would be help us to insert in the same elasticsearch server and solved our problem.
could you help us please?
attached all software version
Centos 7.5
ZabbixServer 4.0
Mysql 5.7.24
Elasticserarch : 6.1.4
I'm Joel Urtubia from Chile, I'm works on a telecom company on Chile, now we have monitoring 100K of devices with elasticseach with 8K values per second, but with this number the zabbix server was full, about the last we need to separate the load in two zabbix servers , ....we was talked with a collegue and read a lot information , and think that have tree ways to solved it:
1- use a zabbix proxy.
2- use an elasticsearch database independent for each zabbix server.
3- use an elasticsearch database but each zabbix server inserted on diferent index name
for solution 1, zabbix proxy use a local database for store the results , and is too slow for growing in devices, we was tested it.
for solution 2, this should be work but the main problems is that all data was separatly , is not scale for views and is difficult development a view for end customer.
for solution 3 , this should be the best solution, but I see that each zabbix only insert in the same index name, and it could be a problem... is posible that open a feature request??, for :
each zabbix server insert with a unique ID or name and will create a index, like this:
zabbixServer1 :
Index name elastic:
zabbixServer1-uint-date
zabbixServer1-dbl-date
zabbixServer1-str-date
zabbixServer1-log-date
zabbixServer1-text-date
zabbixServer2 :
Index name elastic:
zabbixServer2-uint-date
zabbixServer2-dbl-date
zabbixServer2-str-date
zabbixServer2-log-date
zabbixServer2-text-date
it would be help us to insert in the same elasticsearch server and solved our problem.
could you help us please?
attached all software version
Centos 7.5
ZabbixServer 4.0
Mysql 5.7.24
Elasticserarch : 6.1.4