Network monitoring

[Wesleigh]

Hi all,

 

Was appointed to my first NM position in October and have been constantly fighting fires since.

 

9 months in, I’ve found myself in the position to be able to start looking forwards to what’s next, and I know there’s a huge amount I want to implement - and fast.

 

My main priority at this point is to be able to monitor our internal network load (with accountability). I often find that we have seriously slow patches, usually caused by an overly eager device putting unnecessary load on our network. Everything going in and out is monitored to death via our Sophos XG.

 

Does anyone have any recommendations for persistently monitoring an internal network. Looking at about 600 connected devices at any one time.

 

TIA.

[PlantHead]

We use PRTG for our network monitoring.

Very happy with it and has helped us identify a number of cables and switches that needed replacing or redesigning.

https://www.paessler.com/

[mavhc]

What switches do you have?

 

In general update the firmware, enable SNMP (v3 preferably), and install a zabbix appliance vm to collect the data

[TechMonkey]

We use PRTG free (up to 100 nodes) but we aren't monitoring every port, just that things are available and some extras. LibreNMS is meant to be very good.

[networkmangler]

I have a couple of Raspberry Pi's, one running Cacti and the other running NEMS NEMS Linux | Enterprise Asset Monitoring

[dmj]

Prometheus/Grafana stack is considered pretty much de facto standard in SRE circles these days

https://prometheus.io/

https://grafana.com/

 

It's not for the faint hearted, it's quite a high end system.

[Norphy]

Another vote for PRTG. It’s very good value compared to a lot of the other commercial monitoring tools I’ve looked at.

 

If you’ve already got a System Center license for DPM or whatever, SCOM is worth a look but is pretty weak at monitoring switches.

[Wesleigh]

We use PRTG for our network monitoring.

Very happy with it and has helped us identify a number of cables and switches that needed replacing or redesigning.

https://www.paessler.com/

 

Had a look at PRTG, looks very promising. Pricing looks great as well!

 

What switches do you have?

 

In general update the firmware, enable SNMP (v3 preferably), and install a zabbix appliance vm to collect the data

 

I've considered Zabbix a couple of times.

 

Switch Layout:

Core:

2x HP 5500-48G

 

Aggregate Switches:

8x Aruba 2920 48G PoE+

Access Switches:

4x Aruba 2530 48G PoE+

26x Aruba 2530 48G

We use PRTG free (up to 100 nodes) but we aren't monitoring every port, just that things are available and some extras. LibreNMS is meant to be very good.

 

PRTG Free is definitely something I'll check out. Thanks for this.

I have a couple of Raspberry Pi's, one running Cacti and the other running NEMS NEMS Linux | Enterprise Asset Monitoring

I've trialled both of these, however haven't ever considered chucking them onto a Raspberry Pi before...

Prometheus/Grafana stack is considered pretty much de facto standard in SRE circles these days

https://prometheus.io/

https://grafana.com/

It's not for the faint hearted, it's quite a high end system.

It's fair to say that I am probably considered as one of the faint hearted. I've added them to the list to check out but thank you for the heads up.

Another vote for PRTG. It’s very good value compared to a lot of the other commercial monitoring tools I’ve looked at.

 

If you’ve already got a System Center license for DPM or whatever, SCOM is worth a look but is pretty weak at monitoring switches.

I've had a look, and really the switch management is one of my key things. Thanks for the third PRTG vote. It's at the top of my list for now.

[Norphy]

Nedi is a good visualisation tool which has some basic monitoring built into it. It's free: https://nedi.ch

 

WhatsUp Gold is a commercial monitoring system which is also good but fairly expensive.

 

If management is more your thing than monitoring, looking at HP's own tools might be worth considering too. IMC or Airwave.

 

If you're really masochistic, there's also Nagios and its various derivatives but every time I look at Nagios, I end up turning the air blue and various sized children run away in fear so maybe keep away from that one.

[jthompson]

If you're really masochistic, there's also Nagios and its various derivatives but every time I look at Nagios, I end up turning the air blue and various sized children run away in fear so maybe keep away from that one.

 

I'm going to throw in a suggestion for Checkmk (Raw edition is free). It's derived from Nagios but saves you from pretty much all of the infamous Nagios faff. There's quite a high degree of automated discovery in it now and the online/inline documentation of everything is good. At its heart if follows the basic Nagios model of hosts (Up/Down) and their services (OK/Warn/Crit) with graphed performance data for most things.

 

On top of all the usual SNMP stuff, the agents for servers add a tonne of useful stuff (Windows tasks, Veeam jobs, etc.).

Edited June 15, 2021 by jthompson

[supportman]

If you have ubiquity switches you can check the load at any one time on the management interface.

 

If certain clients are causing an issue why not limit their connection speeds in some way? Perhaps on the wireless controller?

[KevinB]

+1 for PRTG. Very easy to use and displays things in a simple dashboard view if that's your bag. (It is mine!)

[TriggerHappyUK]

We use Zabbix (https://www.zabbix.com/) - Client on each device

 

Then we use Grafana (https://grafana.com/) to display it all nicely on a monitor in our office.

[bald_pig]

Purely for network monitoring, I use The Dude:https://mikrotik.com/thedude

For server monitoring, I use PRTG as well.

[danielghares]

+1 for zabbix and grafana. Completely free, and you can do some cool stuff once you dig into it. I'd recommend installing the front end and database on different servers though, as we were having performance issues with them on the same VM.

[cw222]

PRTG (we usually use for more targeted stuff with the sensors rather than an overall picture), Prometheus/Grafana ++, and IMC used for different bits. With Grafana you don't have to re-invent the wheel either and can get started from some preconfigured dashboards. Edited June 18, 2021 by cw222

[slugshead]

 

Switch Layout:

Core:

2x HP 5500-48G

 

Aggregate Switches:

8x Aruba 2920 48G PoE+

Access Switches:

4x Aruba 2530 48G PoE+

26x Aruba 2530 48G

 

 

IMC https://www.hpe.com/us/en/networking/management.html

 

We use it, it's great with our Aruba network. It writes back to switches too so can manage VLANs/port security/Names/IP's etc etc

 

And it gives a good dashboard to have open on the screen so when people walk into the IT office they get to see nerdy stats.

 

 

I do use solarwinds too, but that's another story

[TechMonkey]

IMC https://www.hpe.com/us/en/networking/management.html

 

We use it, it's great with our Aruba network. It writes back to switches too so can manage VLANs/port security/Names/IP's etc etc

 

And it gives a good dashboard to have open on the screen so when people walk into the IT office they get to see nerdy stats.

 

[ATTACH=CONFIG]62264[/ATTACH]

 

I do use solarwinds too, but that's another story

 

Can you give a rough idea how much it cost? I've looked at the HP systems before but presumed that as you have to contact them for a quote they are painfully expensive.