Proxies vs Agents for Distributed Network

[JPhillips]

Hi guys,

I am trying to design a monitoring system for my business. It involves a highly shallow, distributed network, made up of 200+ locations, each with a router, Raspberry Pi, and 1-4 other network devices. I have control over most of the routers, but I want to set it up so at the very least I can monitor the sites as discrete entities. My first thought was to set up each Pi as a Zabbix proxy, run detection on the other devices, and report that way. However, seeing how proxies are treated in the software, that would make configuration a nightmare.

So, a few questions. Can I use agents instead of proxies, and then use them to collect information about the Pi's and the other network devices and send it back to the server over the internet, or do I need to use proxies for that? And if I do, is there any way to make handling that many proxies manageable?

Thanks in advance

[LenR]

You probably could do it via the agent, but it would probably require more "roll your own" code. If it saves you the configuration of 200+ proxies, it might be a net gain. We had a pre Zabbix database of network devices. We scripted adding and removing devices from Zabbix based on that database using the API. It works much better than network discovery because a lot of location, host grouping and device info already existed in the db.

[max.ch.88]

You don't need any proxies or agents. SNMP agents, which are usually present in network devices, will be enough to monitor devices.

[JPhillips]

Thanks for the input Len and Max.

The setup isn't a problem, I need to dial in to the sites and perform full maintenance anyway. The biggest problem I've found is the fact that whenever you assign things to proxies it's always via dropdown. For a few proxies that's fine, but when you have to search through 200+ items in a dropdown it's less than ideal

Regarding SNMP, I could do that, but I'm not sure I like the idea of sending the SNMP data over the internet. I would prefer to send the SNMP data to something local and send it via an encrypted stream

The other thing I want to know is if it is possible to use the agent to monitor other items on the same network. For example, could I install an agent on a completely remote network with no open ports and use it to monitor other SNMP devices on that network?

[max.ch.88]

Between head office and branches use VPN usually. You can sending the SNMP data over VPN and have no problem with security. Or use SNMPv3 and ACL. Connection the branches to office has many network solutions.
"...use it to monitor other SNMP devices..." - yes, you can.

[JPhillips]

Thanks Max.

I can't rely on there being a VPN connection to the sites at this stage, so secured Zabbix connections seem like the next best option.

For monitoring SNMP devices using the agent, can you point me in a direction of how to set it up? I assume I have to add SNMP commands to the template I use for the agents on those sites?

[max.ch.88]

https://www.zabbix.com/documentation...itemtypes/snmp and Template SNMP Generic for a start