Question #19
What are two reasons that cause late collisions to increment on an Ethernet interface? (Choose two.)
A
when Carrier Sense Multiple Access/Collision Detection is used
B
when one side of the connection is configured for half-duplex
C
when the sending device waits 15 seconds before sending the frame again
D
when a collision occurs after the 32nd byte of a frame has been transmitted
E
when the cable length limits are exceeded
English
This detailed analysis will help you understand the question and the underlying networking concepts.
---
### Question Analysis: What causes late collisions?
The question asks for **two reasons** that cause **late collisions** to increment on an Ethernet interface. This immediately points us towards understanding what "late collisions" are and under what circumstances they occur.
#### Key Concepts for Beginners:
1. **Ethernet:** A common technology for local area networks (LANs). In its early forms, it was a shared medium.
2. **Half-Duplex vs. Full-Duplex:**
* **Half-Duplex:** Devices can either send *or* receive at any given time, but not both simultaneously. It's like talking on a walkie-talkie – only one person can speak at a time. This mode uses a mechanism called CSMA/CD.
* **Full-Duplex:** Devices can send *and* receive simultaneously. It's like a phone conversation – both parties can talk at the same time. Full-duplex connections do *not* experience collisions.
3. **Collision Domain:** A network segment where only one device can transmit at a time. If two devices transmit simultaneously, a collision occurs. Half-duplex environments operate within a collision domain.
4. **CSMA/CD (Carrier Sense Multiple Access/Collision Detection):** This is the access method used in half-duplex Ethernet to manage shared access to the network medium and handle collisions.
* **Carrier Sense:** Before transmitting, a device "listens" to the network to see if it's currently in use (if there's a "carrier" signal). If the line is busy, it waits.
* **Multiple Access:** Multiple devices can access the shared medium.
* **Collision Detection:** If two devices transmit at the exact same time (after sensing the line was clear), their signals collide. When a device detects a collision, it immediately stops transmitting, sends a "jam signal" to ensure all other devices know a collision occurred, and then waits a random amount of time (backoff period) before attempting to retransmit.
5. **Slot Time:** A critical concept in Ethernet. It's the minimum time it takes for a signal to propagate across the maximum length of an Ethernet segment and back (a "round-trip delay"). This time dictates the minimum frame size (64 bytes) for Ethernet. A transmitting station *must* be able to detect a collision within this slot time. If a collision is not detected within this time, it becomes a "late collision."
6. **Normal Collision:** Occurs within the first 64 bytes (the "slot time") of a frame transmission. This is an expected event in half-duplex Ethernet, and CSMA/CD handles it gracefully.
7. **Late Collision:** A collision that occurs *after* the first 64 bytes of a frame have been transmitted. This is problematic because it means the CSMA/CD mechanism failed to detect the collision in time. The transmitting station, having already sent more than 64 bytes, might assume it successfully transmitted a valid frame, even though a collision corrupted it. Late collisions indicate an underlying network problem.
---
#### Detailed Option Analysis:
Let's evaluate each option based on our understanding of late collisions:
* **- when Carrier Sense Multiple Access/Collision Detection is used**
* **Explanation:** CSMA/CD is the *mechanism* used to prevent and detect collisions in half-duplex environments. Normal collisions are a part of CSMA/CD's operation. However, late collisions happen precisely because CSMA/CD *fails* to detect the collision within its allowed time window (the slot time). So, while late collisions *occur* in environments where CSMA/CD is used, CSMA/CD itself is not the *cause* of the *lateness*. It's the context, not the reason for the specific "late" characteristic.
* **Verdict: Incorrect.**
* **- when one side of the connection is configured for half-duplex**
* **Explanation:** This scenario often describes a **duplex mismatch**. Imagine one device (Device A) is configured for full-duplex (sends and receives simultaneously, no collision detection) and the other device (Device B) is configured for half-duplex (sends or receives, uses CSMA/CD).
* Device A (full-duplex) assumes it has a dedicated link and can transmit whenever it wants without listening for others or detecting collisions.
* Device B (half-duplex) follows CSMA/CD rules. It listens before transmitting.
* If Device A is transmitting, Device B will sense the carrier and wait.
* However, if Device B starts transmitting, and then Device A *also* starts transmitting (because Device A doesn't perform collision detection and doesn't know Device B is sending), a collision will occur.
* Crucially, because Device A isn't detecting collisions, it won't stop transmitting. This means Device B's signals will collide with Device A's for an extended period, potentially *after* Device B has already sent a significant part of its frame (beyond the 64-byte slot time). Device B will then record a late collision.
* **Verdict: Correct.** Duplex mismatch (specifically one side half-duplex and the other full-duplex) is a classic cause of late collisions.
* **- when the sending device waits 15 seconds before sending the frame again**
* **Explanation:** This option describes a very long, fixed retransmission delay. In CSMA/CD, the backoff period after a collision is random and much shorter (typically in milliseconds or microseconds), not a fixed 15 seconds. This is not how Ethernet handles retransmissions after a collision.
* **Verdict: Incorrect.**
* **- when a collision occurs after the 32nd byte of a frame has been transmitted**
* **Explanation:** This option attempts to define a late collision. A late collision is generally defined as one that occurs *after* the 64th byte (the slot time) of a frame has been transmitted. If a collision occurs after the 32nd byte but before the 64th byte, it would still be considered a *normal* collision because it falls within the slot time where detection is expected. While this option points to the *characteristic* of a late collision (occurring later in the frame), it's not a *cause* for it, nor is the 32nd byte the definitive threshold for "late." The question asks for *reasons that cause* late collisions, not a definition of *when* a collision is late.
* **Verdict: Incorrect.**
* **- when the cable length limits are exceeded**
* **Explanation:** Ethernet standards specify maximum cable lengths (e.g., 100 meters for 100BASE-TX). These limits are in place to ensure that the "slot time" calculation (the round-trip delay for a signal) remains valid. If a cable is too long, the electrical signals take too much time to travel from one end to the other and back.
* A transmitting station might send a frame.
* A collision might occur at the far end of the excessively long cable.
* The collision signal travels back towards the transmitting station.
* However, because the cable is too long, the collision signal arrives *after* the transmitting station has already finished sending the first 64 bytes (the slot time) of its frame.
* When the collision signal finally arrives, the station records it as a late collision because it wasn't detected within the expected window.
* **Verdict: Correct.** Excessive cable length directly prevents timely collision detection, leading to late collisions.
---
### Conclusion:
The two primary reasons that cause late collisions on an Ethernet interface are:
1. **When one side of the connection is configured for half-duplex** (implying a duplex mismatch with the other side being full-duplex).
2. **When the cable length limits are exceeded.**
These are classic troubleshooting scenarios in older or misconfigured Ethernet networks, especially those operating in half-duplex mode. Modern switched networks primarily use full-duplex, where collisions (and thus late collisions) do not occur.
日本語
シスコ認定資格(CCNA)試験対策として、イーサネットインターフェースにおける「レイトコリジョン(Late Collisions)」が発生する主な原因について、プロのネットワークエンジニアの視点から詳細に解説します。
---
# 試験問題:レイトコリジョンの発生原因
**問題:**
イーサネットインターフェースにおいて、レイトコリジョン(Late Collisions)のカウントが増加する原因となる2つの理由はどれか?(2つ選択)
---
## 1. 正解 (Correct Answers)
1. **デュプレックスの不一致 (Duplex Mismatch)**
2. **規定を超えるケーブル長 (Cable length limits exceeded)**
---
## 2. 専門用語の定義
解説の前に、まずは「コリジョン」と「レイトコリジョン」の違いを整理します。
* **通常のコリジョン (Normal Collision):**
イーサネット(半二重)において、フレームの最初の64バイト(512ビット)を送信している間に発生する衝突です。これは半二重環境では正常な動作の一部として処理されます。
* **レイトコリジョン (Late Collision):**
フレームの最初の**64バイト(512ビット)を送信し終えた後**に検出される衝突のことです。これはネットワーク上の異常(エラー)としてカウントされます。
---
## 3. 詳細解説
### 原因①:デュプレックスの不一致 (Duplex Mismatch)
これが実務および試験において最も一般的な原因です。
* **メカニズム:**
接続された2つの機器のうち、一方が「フルデュプレックス(全二重)」、もう一方が「ハーフデュプレックス(半二重)」に設定されている場合に発生します。
* **なぜレイトコリジョンになるのか:**
フルデュプレックス側のポートは、送信と受信を同時に行えるため、CSMA/CD(衝突検出)を行わずにデータを送信し続けます。一方、ハーフデュプレックス側が送信中にフルデュプレックス側からのデータを受信すると、ハーフ側はそれを「衝突」とみなします。このタイミングがフレームの64バイト以降であった場合、レイトコリジョンとして記録されます。
### 原因②:規定を超えるケーブル長 (Cable length limits exceeded)
物理レイヤ(レイヤ1)の設計ミスが原因です。
* **メカニズム:**
イーサネット(10Base-T/100Base-TXなど)には、衝突を正しく検出するために「スロットタイム」という概念があり、最大ケーブル長(通常100メートル)が定められています。
* **なぜレイトコリジョンになるのか:**
ケーブルが長すぎると、電気信号の伝搬遅延(Propagation Delay)が大きくなります。送信側がフレームの64バイト分を送り終える前に、衝突信号(ジャム信号)が戻ってこなければなりませんが、ケーブルが長いと衝突信号が戻るまでに時間がかかり、64バイト送信後に到着してしまいます。これがレイトコリジョンです。
---
## 4. その他の補足事項(トラブルシューティング)
エンジニアとして、レイトコリジョンが発生しているインターフェースを確認した際は、以下のコマンドで状況を把握します。
```bash
Router# show interfaces GigabitEthernet 0/1
...
0 errors, 0 collisions, 152 late collisions <-- ここを確認
...
```
### 対策
1. **デュプレックス設定の確認:** 両端の機器を `Auto-negotiation` にするか、両端を明示的に `Full` に固定します。
2. **物理配線の確認:** ケーブルが100mを超えていないか、または劣化したケーブルや不良なトランシーバ(SFP等)が原因でないかを確認します。
---
## まとめ
CCNA試験において、**「Late Collision = デュプレックス不一致 または ケーブル長超過」**という組み合わせは非常に重要です。通常のコリジョンは半二重環境の仕様ですが、レイトコリジョンは「ネットワークの不具合」であることを確実に理解しておきましょう。
繁體中文
這是一題經典的 **CCNA (Cisco Certified Network Associate)** 考試題目,主要考驗考生對於乙太網路(Ethernet)物理層與資料連結層運作機制,特別是 CSMA/CD 協定與雙工模式(Duplex)的理解。
以下是以專業網路工程師視角進行的詳細解析:
---
### 考題分析:What are two reasons that cause late collisions to increment on an Ethernet interface? (Choose two.)
#### 正確答案:
1. **Duplex Mismatch (雙工模式不匹配)**
2. **Cable length limits are exceeded (線纜長度超過標準限制)**
---
### 詳細技術解析
在深入了解原因之前,我們必須先定義什麼是 **Late Collision(延遲碰撞)**:
在標準的半雙工(Half-Duplex)乙太網路中,碰撞應該發生在影格(Frame)傳輸的前 **64 位元組(Bytes)** 之內。如果碰撞發生在傳送完 64 位元組之後,就被稱為「Late Collision」。
#### 1. 雙工模式不匹配 (Duplex Mismatch) —— 最常見的原因
這是現代網路環境中最常導致 Late Collision 的原因。
* **情境**:當鏈路的一端設定為 **Full-Duplex(全雙工)**,而另一端設定為 **Half-Duplex(半雙工)**。
* **原理**:
* 全雙工端不會監測碰撞(CSMA/CD 關閉),它認為自己可以隨時傳送與接收資料。
* 半雙工端則會遵循 CSMA/CD 協定。
* 當全雙工端在半雙工端正在傳送資料的中途(且已超過 64 Bytes)突然傳送資料時,半雙工端會偵測到碰撞。由於此時半雙工端已經傳送了超過 64 Bytes,它會記錄一次 **Late Collision**。
* **影響**:會導致嚴重的網路效能下降、封包遺失與應用程式延遲。
#### 2. 線纜長度超過標準限制 (Cable Length Exceeded) —— 物理層限制
乙太網路的時序(Timing)是根據線纜的最大容許長度設計的。
* **原理**:
* 乙太網路定義了一個「槽時間(Slot Time)」,這是訊號在網路上往返傳播所需的最長時間。
* 在 10/100 Mbps 環境中,如果線纜過長(例如超過了 Cat5e/Cat6 標準的 100 公尺),訊號傳播的延遲會增加。
* 傳送端可能已經傳送完了前 64 Bytes,但碰撞訊號因為距離太遠,還沒傳回到傳送端。等傳送端收到碰撞訊號時,已經超過了處理正常碰撞的時間視窗,這就會被歸類為 **Late Collision**。
---
### 補充知識:為什麼是 64 位元組?
在乙太網路標準中,64 Bytes 是最小影格長度。這個長度的設計是為了確保在網路的最遠兩端發生碰撞時,傳送端在完成傳送之前,一定能偵測到碰撞訊號(Collision Detection)。這被稱為 **Collision Window(碰撞視窗)**。
### 工程師的排錯建議 (Troubleshooting)
當你在 Cisco Switch 上執行 `show interfaces` 指令,發現 `late collisions` 計數器持續增加時,請依序檢查:
1. **檢查雙工設定**:
* 使用 `show interfaces status` 查看兩端介面的 Speed 與 Duplex。
* **金律**:兩端必須一致。建議統一設定為 `Auto/Auto`,或者兩端皆手動強制設定(例如 `100/Full`)。
2. **檢查物理線路**:
* 確認線纜長度是否符合 TIA/EIA-568 標準(通常小於 100 公尺)。
* 檢查是否有劣質轉接頭或損壞的線纜導致訊號衰減與延遲。
### 總結
**Late Collision** 通常代表網路存在**設定錯誤(雙工不匹配)**或**硬體違規(線路過長)**。在全雙工的現代交換式網路中,Late Collision 的數值應該維持在 **0**。如果出現此數值,通常優先排查雙工模式設定。
简体中文
这是一份针对 CCNA 认证考试中**延迟冲突(Late Collisions)**问题的深度技术解析。
---
# CCNA 考题解析:以太网接口延迟冲突(Late Collisions)
### 1. 题目回顾
**问题:** 导致以太网接口上延迟冲突(Late Collisions)计数增加的两个原因是什么?(选择两项)
**正确选项通常包括:**
1. **双工不匹配 (Duplex Mismatch)**
2. **电缆长度超过限制 (Cable distance limits exceeded / Propagation delay)**
---
### 2. 核心技术概念:什么是延迟冲突?
在深入分析原因之前,我们必须理解以太网的标准机制:
* **正常冲突(Normal Collision):** 在半双工以太网中,冲突应发生在数据包的前 **64 字节(512 比特)**之内。这是以太网槽时间(Slot Time)的定义。
* **延迟冲突(Late Collision):** 指冲突发生在帧的**第 64 个字节之后**。
* **CSMA/CD 限制:** 在标准的 CSMA/CD(载波侦听多路访问/冲突检测)协议中,一旦发送方发送了超过 64 字节而未检测到冲突,它就会认为该帧已成功占用了信道,不再继续监听冲突。如果此时发生了冲突,网卡硬件无法自动重传,该帧会被丢弃。
---
### 3. 原因深度分析
#### 原因一:双工不匹配 (Duplex Mismatch)
这是现网中最常见的延迟冲突原因。
* **技术细节:**
* 当链路的一端配置为 **全双工 (Full-Duplex)**,而另一端配置为 **半双工 (Half-Duplex)** 时。
* 全双工端会随时发送数据,而不进行冲突检测。
* 半双工端在发送数据时会监听冲突。如果全双工端在半双工端已经发送了 64 字节数据后才开始发送,半双工端会检测到一个冲突。
* 由于这个冲突发生在 64 字节之后,半双工端会将其记录为 **Late Collision**。
* **现象:** 全双工端通常会记录 FCS 错误或 CRC 错误,而半双工端则会记录 Late Collisions。
#### 原因二:电缆长度超过限制 (Excessive Cable Length)
这涉及到物理层的传播延迟(Propagation Delay)。
* **技术细节:**
* 以太网设计要求信号在电缆中往返的时间必须短于发送 512 比特数据所需的时间。
* 对于 10/100 Mbps 以太网,双绞线的最大长度限制为 **100 米**。
* 如果电缆过长(例如达到 150 米或更多),信号从一端传到另一端并发生冲突后再传回来的时间,会超过发送前 64 字节的时间。
* **结果:** 发送方在检测到冲突信号时,已经发送完了前 64 字节,从而触发延迟冲突计数。
* **相关因素:** 级联过多的中继器(Repeater)也会增加传播延迟,导致同样的效果。
---
### 4. 网络工程师的排障建议
作为一名 Cisco 网络工程师,当你在 `show interfaces` 输出中看到 `late collision` 增加时,应采取以下步骤:
1. **检查双工状态:**
* 使用 `show interfaces ` 命令。
* 确认两端是否都是 `Full-duplex`。在现代交换网络中,除了连接极老旧的集线器(Hub),所有链路都应为全双工。
* **注意:** 如果一端是 `Auto`,另一端是 `Fixed Full`,由于自动协商机制的限制,`Auto` 端可能会回退到 `Half-duplex`,导致不匹配。
2. **物理链路测试:**
* 如果双工配置正确(均为 Full),则检查物理线路。
* 确认网线长度是否符合 TIA/EIA 568 标准(不超过 100 米)。
* 检查是否存在劣质网线或中间接头过多导致阻抗不匹配。
3. **影响评估:**
* 延迟冲突会导致严重的网络性能下降,因为 TCP 会因为丢包而频繁进入拥塞控制,且以太网层无法自动重传这些损坏的帧,必须由上层协议负责重传。
---
### 5. 总结
**延迟冲突(Late Collisions)** 是半双工逻辑下的异常产物。在现代全双工交换网络中,它们不应该出现。一旦出现,首要怀疑对象是**双工配置不一致**,其次是**物理链路物理规格违规**。
**CCNA 考点记忆:**
* **64 字节后**发生的冲突 = Late Collision。
* **关键词:** Duplex Mismatch(双工不匹配)、Cable Length(电缆长度)。