What is your methodology for diagnosing slow network speeds ?

[Spiky_Gladiator]

Hi,

I'm fairly new to OPNSense and was wondering how you diagnose slow network speeds to pinpoint what's at fault, like is it hardware or software issue, is it a switch or access point itself etc. I currently don't have any issues but things are bound to happen eventually as not all hardware can last decades so I thought I would ask.

So far, I have:

• I heard of a tool called iPerf but haven't had time to check it out yet.
• Checking Speed Tests online that check the connection of the WAN.

Are there any more things ? If so, share !

Thanks

[cookiemonster]

iperf is one of the go-to tools at the admin's disposal and there are ways to use it. For instance using multiple connections when testing.
Otherwise is as usual a case of methodically going by elimination process.
Also speed tests are best through the firewall i.e. doing them from a client on the LAN. Also, some speed test servers behave differently to others.
Last thought for now is that realtek NICs are totally pants.

[Spiky_Gladiator]

Anyone else ?

[Seimus]

Ask your self, does this issues happen

A. from LAN-to-WAN
B. from LAN-to-LAN
C. Impacts both
D. How does the issues present itself; latency, slow throughput, connectivity disruptions
E. When does this issues happens

Know your hardware and applications

A. What hardware do you run your network
B. What is the design of the network
C. What features are you running on your FW/Router/Switch
D. What traffic pattern is affected? General browsing, Gaming latency, VOIP?
E. What is the MTU set on your FW/Router/Switch and devices?

Network performance issues step-by-step

1. Check All interfaces from source to destination IP
2. Any errors?
3. Any interface flapping?
4. What is the BW utilization for the interfaces during the issues?
5. What is the CPU utilization for the Network devices in the path from source to destination?
7. What is the state of BUM traffic, any STP loops, and BUM floods?
8. What routing protocols are you using, are they stable, is there any re-convergence happening?

Tools to test and play with

1. Ping - ping from the source to destination and destination to source, what is the latency, packet loss?
2. Trace - trace from source to destination and destination to source, is the path symmetrical?
3. Trace + Ping - identify the HOPs from source to destination and ping each of them, does any HOP show latency, packet loss?
4. IPerf3 - set two endpoints for iperf3 client and a server IntraVLAN(not routed thru a GW) and InterVLAN(routed thru a GW)
5. Speedtest - run speedtest like from Ookla to see what is your WAN BW throughput, does it reach desired contracted BW?
6. Monitoring - set Uptime-kuma or Smokeping set destination to certain local always online nodes and remote nodes - which of them show problem?
 a. Only remote? - Potential WAN issues (ISP)
 b. Only local?  - Potential local issues (errors, flaps, CPU spikes, BW saturation, etc.)
 c. Both?        - Potential local issues (errors, flaps, CPU spikes, BW saturation, etc.)


T-shooting a performance issue is about to set a timeline and proper description of what the issues is and when it happens. After that you can methodically check from the lowest levels.