It's Not the Network, It Pings Marty!
summary
The talk explains why receiving ICMP replies (“it pings”) doesn’t guarantee application connectivity: missing routes or absent return paths can let ping succeed while TCP (e.g., curl to port 80) fails. Live diagnostics demonstrated using ping, traceroute, ip route, netstat (showing no process on port 80), DHCP/DNS checks and emphasis on verifying the routing table and return path; performance testing examples used sockperf (about 2.5 ms RTT and ~880 Mbps over 20 s) and nttcp on Windows, and the recommendation is to use FQDNs rather than raw IPs.
AI-written summary · about
chapters (11)
- 0:00Network intro
- 1:03Network fundamentals
- 5:57Network failure handling
- 9:11Local network IP
- 12:22OSI model protocols
- 14:30Private vs Internet
- 18:45Latency bandwidth throughput
- 20:08Network troubleshooting
- 26:12Network test tools
- 28:02DNS resolution best practices
- 30:42Q&A
related talks
34mIt's Not Always the Network (But Here's How to Prove It): Kubernetes Packet...
15mHostping: Diagnosing Intra-host Network Bottlenecks in RDMA Servers
29mNetwork Fault Finding System: Packet Loss Triangulation
36mPeeling Away the Layers of the Network Stack
32mActive Fault Finding in Networks
53mHow shit works: TCP/IP
36mNetwork Machinery: A United-Front for Network Troubleshooting with CRDs
46mCurio Collection of Web Protocols