I’m new here and basically new to the product.
we want to start visualizing our network devices metrics (SNMP) using Grafana.
searching the web I’ve understood that Telegraf is a very recommended agent due to a vast number of inputs/outputs plugins and for being extremely lightweight.
we want to use a centralized DB in AWS while having multiple on-premise agents polling data from devices via SNMP and sending it to remote DB.
from what I’ve found I saw only installations with local DB (influxDB) and none of remote DB.
do you have any recommendations and document that we can use to support that architecture?
Grafana does not care where your data store is.
Provided you have good bandwidth access from Garafana to the data store so
that queries and results can be transferred efficiently, you can put your data
store on another machine on your local network, another machine at a data
centre, or hosted somewhere “in the cloud” (provided you have an IP address to
If the data store is anywhere outside your directly-controlled network/s, I
strongly recommend ensuring that the connection is secured with a VPN or TLS
When you configure a data store in Grafana, you tell Grafana where to find it.
If it’s on the local machine you would commonly specify the address as
127.0.0.1 or localhost, but if it’s elsewhere all you need to do is specify
the IP address or DNS-resolvable hostname of the machine running the data
store and Grafana will happily use it.
Obviously any intermediate firewalls need to allow access from the Grafana
server to the data store so that queries can be performed.
I hope that helps, and have fun with Grafana
I should add that telegraf does not care where your InfluxDB instance is
In the [[outputs.influxdb]] stanza of your telegraf configuration file, you
simply specify the location of InfluxDB using the “urls” directive.
The default is localhost, but you can specify any IP address or DNS-resolvable
thanks for replaying
so if my goal is a centralized DB for all my sites
would you recommend using on-primes telegram agents sending data to cloud influxDB?
or there a better best practices recommended?
Separate agents all writing to a centralised DB sounds like a good solution to
Bear in mind that any network connectivity problems will result in data not
being written to the centralised DB, so if you do not have a local DB keeping
a copy, that data will be lost.
Depending on what you have for network connections, though, this may not be a
got it, I didn’t consider this option so it sounds that a local copy is important.