Introduction to SNMP

This topic is to discuss the following lesson:

https://networklessons.com/cisco/ccie-routing-switching/introduction-to-snmp/

Great lesson! I had to implement an SNMP change control recently. Nice to have a great refresher.

good explaination

Fahad ALFadani
KSA

The tip for Observium is really helpful, thanks!

Hi Wilfried,

You might also like LibreNMS.

It’s the exact same thing as Observium but it’s free. It supports instant updates and alerting out of the box. For Observium you’ll have to pay when you want automatic updates and alerting.

Rene

Hi Rene,
What is the difference between SNMP get and get next?

Hi Ananth,

That’s a good question.

We have already seen in this lesson the usage of SNMP GET sent from the SNMP Manager is to gather a specific object instance from the SNMP agent.

Or The GETNEXT command requests a list of instances from a remote entity, but expects the next variable in the tree back. So meaning If a GETNEXT is issued on an object then the agent MUST return the next instance in the MIB tree.

Does it make sense?

Under SNMPv3 should it be “noAuthnoPriv” instead of “noAutonoPriv”? :smiley:

1 Like

Hello Chris

Thanks for catching that, I’ll let @ReneMolenaar know.

Laz

Hi Rene,

Need to confrm if we have configure SNMPV3 on NMS and we are configuring SNMPV2 config on router, Whether it will work.

Hi Owais,

The different SNMP versions are not compatible with each other but if you have a NMS that supports SNMPv3, there’s a 99% chance it also supports SNMPv2.

Rene

Rene, my greetings!

There is a little mistake, SNMP version 2 introduced the notion of an INFORM

SNMP version “3” deals with this problem with an alternative message which uses an acknowledgment called the inform message.

Best regards,
Aleksei

Hello Aleksei,

You are right, I’ll update and fix this. Thanks for sharing!

Rene

Hi guys,

Do you know how how connect grafana with observium?
Or even can you provide a guide to setup grafana for networking devices?

Thanks

Hello Giovanni

We don’t currently have configuration examples or lessons that include Grafana or Observium. However, you can find some information and step by step procedures about Grafana and how to use it for various Cisco devices. Some example links can be found below:


Grafana on Cisco.com
Observium on Cisco.com

I hope this has been helpful!

Laz

Another basic question…
If we want to see the total bandwith usage in one interface in a specific time unit, we should check the maximum bw on ingress OR on egress of that interface at that time.
I understand this, but Im not sure why we dont summ the in+out traffic to get the total bandwith in a interface in a specific time unit (green color + blue color )

For example, in a traffic section on a switch interface on observium how can I read the values like these.

Now:
.
.
.
.
.
.
Rx: 16.67G
Tx: 16.51G
Agg 33.18G

The actual bw is 16.67 or 33?

Hello Giovanni

It really depends on what you want to measure. Remember that the bandwidth of a particular interface is given for one direction of traffic flow. So a FastEthernet interface has a bandwidth of 100 Mbps in each direction and a GigabitEthernet interface has a bandwidth of 1000Mbps in each direction (assuming full duplex which is the default).

Now the output that you show above gives you the average bandwidth usage for ingress traffic is 16.67G, and the average bandwidth usage for egress traffic is 16.51G. The Agg value is simply the aggregate of both directions. (Rx + Tx)

The question is, what information is important to you? The Rx and Tx values individually are important especially when determining if there are any bottle necks at that particular interface, and in which direction the bottleneck occurs. The aggregate value is important especially when you want to evaluate the total volume of traffic that you network carries over time.

I hope I’ve answered your question. If not feel free to indicate any points that may not be clear so I can further clarify.

I hope this has been helpful!

Laz

Thank you,
If I want to know if there are any bottle neck I should check Tx avarage OR Rx avarage in a specific time, not both is it correct?
For example if I have a switch with 2 uplink of 10G
The avg of tx or Rx must be less than 20G to avoid congestion right? It is not useful for trouble shooting have the aggregate value

Hello Giovanni

Yes, that is correct.

Keep in mind however that you have to look at each direction of traffic on each interface individually. For example, you may have two 10G uplinks and they are configured using Etherchannel, and you measure a total of Tx 14G traffic on the Etherchannel interface over a particular period of time, this doesn’t prove you don’t have a bottleneck. You may have 10Gbps on one interface, and having packets dropped, and 4G on the other. For this reason, individual physical interfaces must be examined.

Also, if you want to take a look at bottlenecks, the average bandwidth usage may not be the best way to determine this, because will always be lower than any peaks in traffic that you may have. I know I mentioned bottlenecks, but now that I think about it, average traffic over time will simply give you an indication of the kind of traffic you can expect, it may not immediately indicate that you have a problem. Also, the larger the time period you use to measure the average, the less likely you are to discover congestion, which typically is intermittant.

Yes, the aggregate value is even less useful, as you don’t know how much of that was in each direction. Now having said all of this, the best way to discover congestion is if you have packet drops due to full egress queues over long periods of time. The average traffic may lead you to further investigation, but dropped packets is a clear indication of this.

I hope this has been helpful!

Laz

Thank you thank you and thank you, I Know that are basic topics but I have to be clear this.

The last one thing ( maby :slight_smile: ) if We have a 10 Gb link fullduplex, that link can carry 10tx+10rx right?