IP Routing & Routing Tables

IP routing is the fundamental mechanism that moves packets across networks to reach their destination.

The Routing Problem

Question: How does a packet get from a server in New York to a server in Tokyo?

Answer: Through a series of routers, each making a forwarding decision based on routing tables.

Routing Table Basics

A routing table is a list of rules that tells the router: "If destination is X, send packet via Y."

View routing table (Linux):

route -n
# or
ip route show
 
# Example output:
# Kernel IP routing table
# Destination     Gateway         Genmask      Flags Metric Iface
# 0.0.0.0         192.168.1.1     0.0.0.0      UG    100    eth0
# 192.168.1.0     0.0.0.0         255.255.255.0 U     0      eth0
# 127.0.0.0       0.0.0.0         255.255.255.0 U     256    lo

Routing Table Columns

Routing Decision Process

Scenario: Packet destined for 8.8.8.8 arrives at your router

Step 1: Check Destination

Is 8.8.8.8 in my routing table?

Step 2: Longest Prefix Match Find the most specific matching route:

Route 1: 8.0.0.0/8 via 192.168.1.1
Route 2: 8.8.0.0/16 via 10.0.0.1
Route 3: 8.8.8.0/24 via 10.1.0.1
Route 4: 0.0.0.0/0 via 192.168.1.1 (default)

For destination 8.8.8.8:
✓ Route 1: matches (prefix /8)
✓ Route 2: matches (prefix /16)
✓ Route 3: matches (prefix /24) ← MOST SPECIFIC
✗ Route 4: matches but less specific

Decision: Use Route 3

Step 3: Forward

Add packet to queue for next-hop router 10.1.0.1
Use interface eth1 (from routing table)
Send packet!

Step 4: Next Router If 10.1.0.1 isn't the final destination, it repeats the process with its routing table.

Routing Scopes

Local Route Directly connected network:

Destination: 192.168.1.0/24
Gateway: 0.0.0.0 (on-link)
Interface: eth0
→ Deliver directly on local network (using ARP)

Remote Route Network reachable via gateway:

Destination: 10.0.0.0/8
Gateway: 192.168.1.1 (router IP)
Interface: eth0
→ Send to gateway, gateway forwards further

Default Route Fallback for everything else:

Destination: 0.0.0.0/0 (matches any IP)
Gateway: 192.168.1.1
Interface: eth0
→ "When in doubt, send to gateway"

Common Routes

Loopback:

127.0.0.0/8 → 127.0.0.1 (lo interface)
All local services reach loopback device

Broadcast:

192.168.1.255/32 (depends on network)
Broadcast address of your network

Link-Local:

169.254.0.0/16 → auto-assigned if DHCP fails
Temporary connectivity between neighbors

Static vs Dynamic Routes

Static Routes Manually configured, doesn't change:

# Add a static route
sudo ip route add 10.0.0.0/8 via 192.168.1.1
 
# Make permanent (Linux)
# Add to /etc/netplan/00-installer-config.yaml or /etc/network/interfaces

Dynamic Routes Learned automatically via routing protocols:

• RIP (Routing Information Protocol)
• OSPF (Open Shortest Path First)
• BGP (Border Gateway Protocol)

Metric: Route Priority

When multiple routes match, metric determines priority (lower = better):

Route 1: 10.0.0.0/8 via 192.168.1.1 metric 100
Route 2: 10.0.0.0/8 via 10.1.0.1  metric 50

Decision: Use Route 2 (lower metric)

Common Metrics:

• Hop count (number of routers)
• Bandwidth
• Latency
• Reliability
• Load

Multi-Path Routing

Send traffic via multiple paths simultaneously (Equal-Cost Multi-Path - ECMP):

Destination 8.8.8.8:
Route 1: via 192.168.1.1 metric 100
Route 2: via 10.0.0.1   metric 100

Same cost → load balance between both routes
25% traffic on route 1
25% traffic on route 2
(example simplified)

Benefits:

• Higher throughput
• Redundancy — if one path fails, use other
• Better utilization

TTL (Time To Live)

Prevents infinite routing loops:

Packet TTL: 64

Router 1: TTL 64 → TTL 63 → Forward
Router 2: TTL 63 → TTL 62 → Forward
Router 3: TTL 62 → TTL 61 → Forward
...
Router N: TTL 2 → TTL 1 → Forward
Router N+1: TTL 1 → TTL 0 → DROP

ICMP "Time Exceeded" sent back to source

Traceroute Uses TTL:

traceroute google.com
# Sends packets with increasing TTL
# Each router responds with ICMP Time Exceeded
# Shows entire path to destination

Routing Problems

Problem 1: No Route Found

Destination: 10.0.0.50
Routing table has no entry for 10.0.0.0/...
No default route configured

Result: ICMP "Destination Unreachable" → packet dropped

Solution: Add route or configure default gateway

Problem 2: Asymmetric Routing

Outbound path: A → Router 1 → Router 2 → B
Return path: B → Router 3 → Router 4 → A

Speeds differ, latency differs
Firewalls may reject return traffic

Problem 3: Routing Loop

Router A: "10.0.0.0 is via Router B"
Router B: "10.0.0.0 is via Router A"

Packet bounces forever
TTL reaches 0, packet dies
(Lots of wasted bandwidth in the meantime)

Solution: Use dynamic routing protocols that detect loops

Routing in Containers/Kubernetes

Container Network:

Pod IP: 10.244.1.50
Container gets default route: 0.0.0.0/0 → bridge
Bridge gateway forwards to Kubernetes network plugin
CNI (Container Network Interface) routes between nodes

Example:

Pod A (Node 1): 10.244.1.50
Pod B (Node 2): 10.244.2.50

Pod A's routing table:
10.244.1.0/24 → local (eth0)
10.244.0.0/14 → default gateway (CNI plugin)

Pod A sends to 10.244.2.50
→ CNI plugin sees it, knows Node 2 has 10.244.2.0/24
→ Tunnels or routes packet to Node 2
→ Node 2 delivers to Pod B

Viewing Detailed Routes

Linux — Advanced Routing Table View:

# Show all routes with more details
ip route show table all
 
# Show routes added by specific source
ip route show from 192.168.1.100
 
# Show routes to specific prefix
ip route show 10.0.0.0/8
 
# Show route taken for specific destination (where-to)
ip route get 8.8.8.8

Adding and Removing Routes

Add a static route:

# One-time (temporary)
sudo ip route add 10.0.0.0/8 via 192.168.1.1
 
# Permanent (different for each distribution)
# Debian/Ubuntu: /etc/network/interfaces
# RedHat/CentOS: /etc/sysconfig/network-scripts/route-*
# Netplan: /etc/netplan/*.yaml

Remove a route:

sudo ip route del 10.0.0.0/8

Routing Best Practices for DevOps

✓ Always configure default gateway ✓ Document static routes (crucial for debugging) ✓ Use dynamic routing for complex networks ✓ Monitor for asymmetric routes ✓ Test routes with traceroute and mtr ✓ Ensure firewall allows traffic in both directions ✓ Avoid routing loops — use proper routing protocols ✓ Set appropriate TTL for your use case

Key Concepts

• Routing table = rules for forwarding packets
• Longest prefix match = most specific route wins
• Metric = priority; lower is better
• Default route = fallback (0.0.0.0/0)
• Dynamic routes learned via routing protocols (RIP, OSPF, BGP)
• TTL prevents infinite loops
• Traceroute shows path by incrementing TTL
• Every router forwards based on its local routing table
• No routing loop detection needed (TTL handles it)