AppD Archive

Suggested controller setup in a multi-datacentre envrionment

CommunityUser
Splunk Employee
Splunk Employee

Hi,

Is it possible to setup a controller in each of the data centre and have the controller sync with each other rather than agents sending HTTP(S) traffic to a single controller. The main purpose is to reduce cross network traffic and keep them localised where possible.

Looking at the way AppDynamics work - I'm guessing licenses would be a blocking issue. It would be difficult to pre-judge the number of agents in each of the data centre (though the overall count would be static) and hence would not be able to setup two controllers with a set number of licenses. 

Thanks

0 Karma

Arun_Dasetty
Super Champion

Hi,

Currently AD does not support the case where agent sends dynamically data to different controller and controller takes care of data sync between different controller and both controller is running fine.

However we support controller in HA mode where if one controller have crashed and the monitored data is still in sync on other controller and you can get the other controller up to make it active controller so that agent reports data to other controller

let us know if the provided doc link helps: http://docs.appdynamics.com/display/PRO13S/Manage+Controller+High+Availability

Thanks,

Arun

0 Karma

CommunityUser
Splunk Employee
Splunk Employee

Thanks - is there a document which shows average data generated by the agent (assuming running in out of box config) ?

0 Karma

Arun_Dasetty
Super Champion

Hi,

We do not have such docs, Let us know if the following information helps:

- by default amount of data reported by agent depends on following with default settings:

  1) number of business transactions (servlets, strusts requests etc), as agent sends metrics like ART, calls per minutes, slow calls for each BT by particular agent node
 2) number of snapshots reported by agent
 3) in addition jvm and jmx metric specific to server (tomcat, jboss) etc and the associated mbean category metrics that discovered by default (jdbc connection pool datasource , web app context under web container for jmx section etc)

 4) number of backends discovered and the associated metrics
5) number of calls for discovered bts as the snapshot collection varies again

However we confirm the network I/0 added by agent is varies on 500 KB or higher per 5 minutes of time period at max, and this varies on above factors and it could be lower as well, also the agent is expected to not add much overheard to network for the data that is sends to controller.

You can use machine agent to see network I/O but this is machine level data which not be specific to java agent process.

Let us know if that information helps.

Thanks,

Arun

0 Karma
Get Updates on the Splunk Community!

See just what you’ve been missing | Observability tracks at Splunk University

Looking to sharpen your observability skills so you can better understand how to collect and analyze data from ...

Weezer at .conf25? Say it ain’t so!

Hello Splunkers, The countdown to .conf25 is on-and we've just turned up the volume! We're thrilled to ...

How SC4S Makes Suricata Logs Ingestion Simple

Network security monitoring has become increasingly critical for organizations of all sizes. Splunk has ...