Loading learning content...
Before packet switching revolutionized data communications, another store-and-forward technique dominated: message switching. This approach, rooted in telegraph and postal systems, treats each complete message as an indivisible unit—storing it entirely at each intermediate node before forwarding to the next hop.
While message switching has been largely superseded by packet switching for most applications, understanding this paradigm is essential for several reasons: it provides historical context for networking evolution, it remains in use for specific applications (email systems, certain military communications), and its limitations directly motivated the development of packet switching.
This page provides an exhaustive exploration of message switching—its operational principles, historical applications, advantages, critical limitations, and why packet switching ultimately prevailed.
By completing this page, you will: (1) Understand the fundamental operation of message switching, (2) Trace the historical evolution from telegraph systems to modern store-and-forward applications, (3) Analyze the advantages and critical limitations compared to packet switching, (4) Calculate message switching delays, and (5) Identify modern applications where message switching concepts persist.
Message switching is a store-and-forward communication technique where a complete message is transmitted from source to destination through intermediate nodes, with each node receiving, storing, and forwarding the entire message before proceeding.
1. Complete Message as Unit of Transfer
Unlike packet switching, which fragments data into fixed-size or variable-size packets, message switching treats the entire message as an atomic unit. A message may be a single character or megabytes of data—it is forwarded as one piece.
Key Implication: A node cannot begin sending until the entire message has been received. For a 10 MB message on a 1 Mbps link, this means waiting 80 seconds before forwarding can begin—at each hop.
2. Store-and-Forward at Each Node
Each intermediate node (switch, relay station) must:
3. No Fragmentation
Messages are not broken into smaller units for transmission. A 100 KB message occupies the link exclusively while being transmitted, blocking other messages.
4. Queuing of Entire Messages
When multiple messages compete for an outgoing link, entire messages queue—not packets. A long message ahead in the queue completely blocks shorter messages behind it until fully transmitted.
Message switching was also known as 'store-and-forward switching' in early literature. The technique directly evolved from telegraph message relay systems where operators would receive a message at one station, write it down (store), and then re-transmit (forward) on the next leg of the journey. The conceptual parallel is exact.
Message switching has a rich history that predates electronic computers. Understanding this evolution illuminates why packet switching emerged as a superior alternative.
The telegraph network was the first global communication system, and it operated on message switching principles:
Manual Relay Operation:
Torn-Tape Relay: Automated systems used paper tape:
This reduced transcription errors but was still message-based store-and-forward.
Telex (Teleprinter Exchange) formalized message switching for business communications:
Characteristics:
Store-and-Forward Services:
Telex remained in use well into the 1990s, particularly for international business, freight, and legal communications requiring delivery confirmation.
As computers became networked, message switching was the natural starting point:
AUTODIN (Automatic Digital Network): U.S. military message switching network, operational from 1962:
SITA Network (Airline Industry): Société Internationale de Télécommunications Aéronautiques:
Why Message Switching Wasn't Enough:
By the 1960s, interactive computing emerged. Users at terminals expected rapid responses—they couldn't wait minutes while a large message from another user monopolized the network.
Paul Baran's Insight (1964): Baran's RAND Corporation work on survivable communications proposed breaking messages into "message blocks" (later called packets). This allowed:
ARPANET (1969): Directly implemented packet switching, proving its superiority:
ARPANET's success led to IP, the Internet, and the dominance of packet switching.
| Era | Technology | Message Unit | Delay Characteristic |
|---|---|---|---|
| 1850s-1920s | Manual Telegraph Relay | Complete message | Hours to days |
| 1930s-1960s | Automated Telex | Complete message | Minutes to hours |
| 1960s-1970s | Computer Message Switching | Complete message | Seconds to minutes |
| 1970s-Present | Packet Switching | Packets (~1500 bytes) | Milliseconds |
Message switching networks typically offered delivery guarantees—the network stored messages until confirmed received. The Internet's 'best effort' model (no delivery guarantee) was initially controversial. TCP at the transport layer, not IP, provides reliability—a direct application of the end-to-end principle that message switching networks didn't follow.
Let us examine the detailed operational flow of message switching networks.
A message switching message typically contains:
| Component | Description |
|---|---|
| Header | Addressing information, priority, handling codes |
| Body | The actual message content (variable length) |
| Trailer | End-of-message indicator, error checking |
Unlike packets, there is no standard maximum size—messages can range from a few bytes to megabytes.
┌──────────────────────────────────────────────────────────────────────┐
│ Message Switching Node │
├──────────────────────────────────────────────────────────────────────┤
│ │
│ ┌──────────────┐ ┌────────────────┐ ┌───────────────┐ │
│ │ RECEIVE │─────→│ STORE │─────→│ PROCESS │ │
│ │ (complete │ │ (disk/memory) │ │ (parse addr) │ │
│ │ message) │ │ │ │ │ │
│ └──────────────┘ └────────────────┘ └───────┬───────┘ │
│ │ │
│ ▼ │
│ ┌──────────────┐ ┌────────────────┐ ┌───────────────┐ │
│ │ TRANSMIT │←─────│ QUEUE │←─────│ ROUTE │ │
│ │ (complete │ │ (by priority, │ │ (select next │ │
│ │ message) │ │ destination) │ │ hop) │ │
│ └──────────────┘ └────────────────┘ └───────────────┘ │
│ │
└──────────────────────────────────────────────────────────────────────┘
Detailed Flow:
Receive Phase
Store Phase
Process Phase
Route Phase
Queue Phase
Transmit Phase
Message switching nodes required substantial storage—capable of holding all in-transit messages. If a downstream link failed, messages accumulated. Military systems often had tape drives or disk storage measured in megabytes (enormous for the era). This storage requirement was a major operational concern and drove interest in packet switching's smaller, bounded buffers.
Understanding the delay characteristics of message switching versus packet switching illuminates why the latter prevailed for most applications.
For a message of size M bits traversing N hops with link bandwidth R:
$$D_{message} = N \times \frac{M}{R} + N \times d_{prop} + T_{queue}$$
Where:
Critical Observation: The entire message is transmitted at each hop sequentially. No pipelining is possible.
For the same message fragmented into P packets of size L bits:
$$D_{packet} = (N + P - 1) \times \frac{L}{R} + N \times d_{prop} + T_{queue}$$
Key Insight: Pipelining! While the first packet is being forwarded from node 1 to node 2, the second packet can be transmitted from source to node 1. After the first packet traverses the network, subsequent packets follow one transmission delay apart.
Scenario: 1 MB message, 3 hops, 1 Mbps links, negligible propagation delay
Message Switching: $$D_{message} = 3 \times \frac{8,000,000 \text{ bits}}{1,000,000 \text{ bps}} = 3 \times 8 \text{ s} = 24 \text{ seconds}$$
Packet Switching (1000-byte packets, P = 1000 packets): $$D_{packet} = (3 + 1000 - 1) \times \frac{8,000 \text{ bits}}{1,000,000 \text{ bps}} = 1002 \times 0.008 \text{ s} = 8.016 \text{ seconds}$$
Result: Packet switching is 3× faster for this scenario.
The advantage grows with more hops. For 10 hops:
Nearly 10× faster!
| Hops | Message Switching | Packet Switching (1KB packets) | Speed Improvement |
|---|---|---|---|
| 1 | 8.0 s | 8.0 s | 0% (same) |
| 2 | 16.0 s | 8.008 s | ~2× |
| 3 | 24.0 s | 8.016 s | ~3× |
| 5 | 40.0 s | 8.032 s | ~5× |
| 10 | 80.0 s | 8.072 s | ~10× |
| 20 | 160.0 s | 8.152 s | ~20× |
Message Switching:
Time: 0──────8──────16──────24 seconds
├──M──→│ │ │
│ ├──M──→│ │
│ │ ├──M───→│
Hop 1 Hop 2 Hop 3 Arrived
Packet Switching:
Time: 0─────.008────.016────.024─── ... ───8.016 seconds
├P1─→├P1─→├P1─→│ │
│├P2─→├P2─→├P2─→│ │
││├P3─→├P3─→├P3─→│ │
││││ │
│ ... (1000 packets) │
││││├P1000─→├P1000─→├P1000─→│
Hop 1 Hop 2 Hop 3 Arrived
With packet switching, the network operates like a pipeline. Every link is productively transmitting simultaneously (after the first packet traverses each link). With message switching, only one link is active at a time.
For interactive applications (terminal sessions, request-response), message switching is catastrophic:
Packet switching interleaves small packets with large transfers. The 100-character response might only wait behind a few packets, not entire messages.
When ARPANET demonstrated interactive remote login (Telnet) with sub-second response times, it validated packet switching's superiority for interactive computing. Message switching could never achieve this—large message transfers would inevitably block interactive sessions. This single demonstration changed networking history.
Despite its limitations, message switching offers genuine advantages that explain its historical importance and residual applications.
Message switching excels when connectivity is unreliable or intermittent:
Store-and-Forward Buffering:
Delay-Tolerant Networking: Modern DTN (Delay-Tolerant Networking) protocols revive message switching concepts:
Email Systems: SMTP (Simple Mail Transfer Protocol) is fundamentally message switching:
Email is message switching adapted for the Internet age.
For asynchronous, non-interactive communication where guaranteed delivery matters more than latency, message switching remains appropriate. Email, batch file transfers, military message traffic, and delay-tolerant networking all benefit from message switching's reliability-first design.
The limitations of message switching are severe enough that it was abandoned for most applications once packet switching demonstrated its capabilities.
| Criterion | Message Switching | Packet Switching |
|---|---|---|
| Interactive Response | Impossible | Excellent |
| Large File Transfer | Acceptable | Better (pipelining) |
| Buffer Requirements | Unbounded | Bounded (MTU) |
| Link Utilization | ~50% max | ~100% possible |
| Latency | High, variable | Low, predictable |
| Guaranteed Delivery | Inherent (storage) | Requires TCP/protocol |
| Offline Recipients | Inherent | Requires application support |
| Priority Handling | Coarse (message level) | Fine (packet level) |
The rise of interactive computing in the 1960s was the death knell for message switching as a general-purpose network technology:
Time-Sharing Systems:
Remote Terminal Access:
Transaction Processing:
Packet switching solved all these problems by fragmenting data into small units that could be interleaved. ARPANET's demonstration of remote login proved the concept, and within a decade, packet switching dominated new network designs.
By the late 1970s, the choice was clear: packet switching for everything except specific non-interactive, high-reliability applications (military message systems, email). The Internet's success validated this conclusion. Message switching remains a historical curiosity and a niche solution—not a mainstream networking technology.
While pure message switching is rare, its concepts persist in modern systems where its advantages outweigh its latency limitations.
Email is the most widespread message-switching-like system:
Message Switching Characteristics:
Differences from Pure Message Switching:
Enterprise message queues (MQ) implement message switching patterns:
Examples:
Message Switching Characteristics:
Use Cases:
DTN protocols explicitly embrace message switching for extreme environments:
Bundle Protocol (RFC 5050):
Applications:
Military communications retain message switching concepts:
AUTODIN Successor Systems:
Requirements Driving Message Switching:
Some IoT scenarios benefit from message switching:
Intermittent Connectivity:
MQTT QoS Levels:
The message switching paradigm—storing complete messages and forwarding with guaranteed delivery—remains valuable for asynchronous, reliability-first applications. Email, message queues, DTN, and military systems all implement its core concepts. Understanding message switching explains why these systems work the way they do.
We have completed an exhaustive examination of message switching—the store-and-forward paradigm that preceded packet switching and remains relevant in specific contexts. Let us consolidate the key concepts:
What's Next:
With store-and-forward, datagram, virtual circuit, and message switching now understood, we turn to a comprehensive comparison of circuit and packet switching. We'll synthesize everything we've learned to analyze when each paradigm is appropriate and how modern networks combine these approaches.
You have mastered message switching—its historical origins, operational mechanics, delay characteristics, advantages, and why packet switching prevailed. You can now analyze message-based systems like email and message queues through the lens of store-and-forward networking, and articulate why this paradigm persists in specific applications. Next, we'll conduct a comprehensive comparison of circuit and packet switching approaches.