Three of the Top 11 Network Troubleshooting Tips
Guest writing for our blog this week is Tony Fortunato, a senior network performance specialist with The Technology Firm. Tony has more than 20 years’ experience working with network pros on designing, implementing and troubleshooting networks, and he has some recommendations for more effective troubleshooting.
In this post, he reviews a few of his top technical tips for faster, more accurate network troubleshooting and the importance of baselining network performance.
An Introduction to My Top 11 Network Troubleshooting Tips
One thing I know: Whoever came up with the name "the cloud" must have been in marketing.
A network is the exact opposite of a pretty, fluffy cloud. Troubleshooting even a simple problem can involve a nightmare of core switches, pulled fiber, routers, patch panels, and access points. After you've checked all the hardware, you still have to know your way around a Web server and the client machine.
The good news is that I've developed a method for streamlining network troubleshooting after spending 20 years leading design, implementation, and training on networks. Below are few of those tips, listed in no particular order.
#1: Starting with a Good Baseline
It all starts with the baseline. Before problems begin, you really need an idea of what your network infrastructure should look like on a good day. A simple baseline can dramatically shorten your troubleshooting time, but a detailed baseline should be able to explain (or even predict) changes to specific components, or how the network will react to an application.
Some good baselines would be:
- 30 ms average ping response time between 192.168.1.80 and 10.1.10.80
- Application ACME generates 18 Mbps of traffic during account queries
- 6 percent packet loss on WiFi segment 192.168.66.0 with 30 percent load
#11: Detecting Flooded Packets
Remember, you won't be able to accurately judge the results if you wait until there's a problem before you pull a baseline. The baseline is a crucial way to quickly identify flooded packets. It’s so crucial, I've listed it as number 11 on the countdown of my 11 most valuable network tests.
While limited flooding is just a normal part of switching, continuous flooding can severely compromise performance. You'll need to connect a protocol analyzer or network monitor to a customer segment. If you detect constant, one-way traffic between two unicast devices from your port, you'll need to start looking at defaults like the ARP table-aging time.
#5: Instant Network Mapping
Number 5 on my list is an accurate, up-to-the-minute topology map. These maps can quickly become old and unreliable, due to the time and resources it takes to keep them updated. You will save considerable time and frustration by using a network analysis tablet with a real-time discovery engine to generate an instant map of up to 30.000 devices - hosts, phones, switches, routers, access points, servers and whatever else can go wrong. This is essential for troubleshooting unknown networks or systems that change frequently with devices added, removed and modified.
A Deeper Dive
We really can only scratch the surface in this blog. For a deeper dive into the details on all 11 tips, feel free to join me for my network troubleshooting webinar. It's full of best practices, hard data, and examples to help you get to the root cause faster. Another tool that can help you get there is the NETSCOUT network analysis tablet, which can quickly diagnose network vs. application failures, line-rate packet capture and analysis (1G / 10G), quality assessments for VoIP/Video/UC, or even 802.11 WLAN planning and deployment. It's just the thing to brighten up your next "cloudy" day.