34368_week_04

Which of the following is NOT a typical example of a network failure?

• Hardware failures due to component degradation

• Environmental disasters like fires or floods

• Regular software updates

• Human-caused sabotage

Answer explanation

The slide on "Network Failures" (slide 5) lists various types of failures that can occur in networks. These include hardware failures, environmental disasters, and human-caused problems. Regular software updates are not typically considered a failure, but rather a normal part of network maintenance. The other options are all mentioned as potential causes of network failures.

What is a key requirement for network survivability in terms of network connectivity?

• Single homing of nodes

• Dual homing of nodes

• Fully meshed network topology

• Star network topology

Answer explanation

The slide titled "Traffic redirection – Basic requirements" (slide 9) mentions "Dual homing: each node connected by at least two spans" as a basic requirement for network survivability. This ensures that each node has at least two independent paths to the rest of the network, providing redundancy in case of a single failure.

In the context of network protection, what does the term "1+1" refer to?

• A protection scheme where traffic is sent on both working and protection paths simultaneously

• A restoration scheme where one backup path is calculated for each working path

• A shared protection scheme where multiple working paths share one backup path

• A scheme where one working path is protected by one backup path, but traffic is only sent on the working path

Answer explanation

The "Linear Networks: 1+1" slide (slide 24) describes the 1+1 protection scheme. In this scheme, traffic is sent simultaneously on both the working path and the backup path. The receiver then selects which signal to use, providing immediate protection in case of a failure on one of the paths.

What is a key difference between span restoration and path restoration in mesh networks?

• Span restoration is faster but less efficient in resource usage

• Path restoration always provides shorter backup paths

• Span restoration requires more signaling between nodes

• Path restoration can only be used in fully connected mesh networks

Answer explanation

The "Comparison: Span vs. Path vs. Segment" slide (slide 38) compares different restoration methods. Span restoration is noted to have shorter notification times (faster) as the failure is dealt with locally. However, it's also mentioned that span restoration is bound to reuse stubs and may result in longer paths, implying less efficient resource usage compared to path restoration, which has "high resource efficiency" listed as an advantage.

What is a p-cycle in the context of network protection?

• A physical cycle of fiber optic cables in the network

• A preconfigured protection cycle that visits all nodes in the network

• A protocol for cycle detection in network topologies

• A type of failure that affects multiple nodes in a cyclic pattern

Answer explanation

The "p-cycle (Protection cycle)" slide (slide 42) defines a p-cycle as a "Preconfigured protection cycle that visits all nodes in the network." It's described as operating similar to a normal ring but designed to provide efficient protection in mesh networks by combining the speed of ring protection with the efficiency of mesh topologies.

Which two elements does survivability require, according to the article?

• Redundancy and complexity

• Redundancy and diversity

• Diversity and simplicity

• Complexity and diversity

Answer explanation

In Section 2.2.2 of the article, when discussing survivability, the authors state: "In addition to the redundancy required by fault tolerance, survivability requires diversity so that the same fate is unlikely to be shared by parts of the system undergoing correlated failures." This clearly indicates that survivability requires both redundancy and diversity. Redundancy helps in dealing with individual failures, while diversity ensures that correlated failures do not affect all redundant components simultaneously.