AEiE0905 IP switching¶
Core idea¶
- IP switching aims to combine IP routing flexibility with faster forwarding for flow-based traffic.
- Ipsilon IP Switching is a classic early architecture in this area.
- Flow classification is central: packets are grouped into flows so a faster switching path can be created.
Ipsilon IP Switching overview¶
- Initial packets of a flow are routed normally at the IP layer.
- When a flow is recognized as worth switching, a shortcut switched path may be established.
- This reduces repeated route lookup overhead for the same flow.
- It is historically important as a bridge idea between pure IP routing and label-based forwarding.
Flow classification¶
- A flow is a sequence of packets sharing selected header fields and treatment.
- Classification may use source address, destination address, protocol, and port values.
- Flow granularity affects scalability and efficiency.
IP service model¶
- Classical IP service is connectionless, best-effort delivery.
- It does not guarantee delivery, order, delay, or bandwidth by itself.
- Reliability, sequencing, and congestion behavior are mostly handled by higher layers such as TCP.
Layering in IP protocols¶
| Layer context | Typical role |
|---|---|
| Link layer | Frame delivery on local medium |
| Internet layer | Logical addressing and routing with IP |
| Transport layer | End-to-end service via TCP or UDP |
| Application layer | User services |
- IP belongs to the network or internet layer.
- ARP maps IP address to MAC address on a local network.
IP packet structure and IP header¶
Important IPv4 header fields commonly tested:
| Field | Function | Key fact |
|---|---|---|
| Version | IP version number | 4 for IPv4 |
| IHL | Header length | In 32-bit words |
| Total Length | Total packet size | Header plus payload |
| Identification | Fragment reassembly support | Used with fragmentation |
| Flags | Fragmentation control | Includes DF and MF |
| Fragment Offset | Fragment position | In 8-byte units |
| TTL | Hop limit against looping | Decremented by each router |
| Protocol | Upper-layer payload type | TCP, UDP, ICMP etc. |
| Header Checksum | Error check on header only | Recomputed at routers |
| Source Address | Sender IP | 32 bits |
| Destination Address | Receiver IP | 32 bits |
- TTL stands for Time To Live, but in practice it acts as a hop-count limit.
- If TTL reaches zero, the packet is discarded and an ICMP error is usually sent.
Recognition cues for IP switching¶
- "Best effort" indicates IP service model.
- "Hop count limiter" indicates TTL.
- "Header only checksum" indicates IPv4 header checksum, not payload protection.
- "Flow-based shortcut switching" indicates IP switching ideas such as Ipsilon.
Common traps in IP switching¶
- TTL is not a wall-clock timer in normal router behavior.
- IP itself is connectionless; a connection-oriented user experience usually comes from transport or application behavior.
- IPv4 header checksum does not protect transport payload.
- Flow classification is not the same as route aggregation.
One-step examples for IP switching¶
- If a packet starts with TTL 64 and crosses 5 routers, its TTL becomes 59.
- A packet with DF set cannot be fragmented by routers.
- A flow defined by source, destination, protocol, and ports is more specific than one defined by destination alone.
IP-switching revision box¶
IP switching groups packets into flows for faster handling. Classical IP service is connectionless and best effort. TTL is the loop-prevention field and is decremented at each router. Learn the major IPv4 header fields and what they actually protect.