In my previous blog post, I introduced Nautilus, a cross-layer cartography framework designed to map Internet Protocol (IP) links to their corresponding submarine cables.
By helping us understand the interdependencies between the physical and network layers, Nautilus sets the stage for a rigorous analysis of how cable disruptions could affect global Internet connectivity.
In this post, I will discuss Xaminer. This resilience analysis tool uses cross-layer maps from frameworks such as Nautilus to provide insights into how physical cable failures impact multiple layers of the Internet stack.
Why Internet Resilience Matters
The Internet is a critical infrastructure in our interconnected world, and its resilience is crucial for maintaining communications and essential services. The physical cables, which carry most of the data traffic and are highly shared across operators, form the backbone of this critical infrastructure. Thus, any disruption to these cables can have far-reaching consequences, potentially leading to widespread outages and economic impacts across multiple organizations and countries.
The Role of Xaminer
Xaminer aims to assess the cross-layer resilience of the Internet infrastructure and quantify the risk posed by various failure events, such as natural disasters or accidental damages. Xaminer relies on cross-layer maps from frameworks such as Nautilus and failure event models (models that capture the intensity of failure events for a geographic region) to analyze the impact of these failures at varied granularities across layers. This would enable network operators and Internet stakeholders to prioritize and devise effective repairs and deployment strategies to minimize disruptions and improve resilience.
Key features of Xaminer include its ability to perform analysis at varied granularities, ranging from a single cable segment to entire countries and global scale analysis. Its modular design enables easy addition of analysis metrics, cross-layer maps, or failure event models.
Xaminer supports joint cross-layer impact analysis for multiple disasters or events, combining the effects of different events on the infrastructure. Additionally, Xaminer can identify patterns and trends independent of specific events.
Case Studies Using Xaminer
Let’s ‘examine’ a few case studies demonstrating Xaminer’s applicability in assessing Internet resilience under failure scenarios.
Regional Impact Assessment
In this case study, Xaminer is used to assess the impact of regional disasters, in this case earthquakes in Japan and the Pacific Northwest (PNW) and hurricanes in the Caribbean, on the global Internet. Xaminer helps us identify the most impacted infrastructure components and the overall risk globally under these regional failures. This helps network operators and regional governments identify the vulnerable parts of the infrastructure and prioritize reinforcement or rerouting efforts.
Read: Keeping the Internet on Following Natural Disasters
In this example(Figure 1), we can see that the network layer components bear a higher risk in Japan, a region with limited cables. In contrast, the physical layer components face maximum susceptibility in the Caribbean region.
Sea Level Rise – Impact Across Countries
Climate change and rising sea levels impact physical infrastructure, such as submarine cable landing stations. Using Xaminer, we generated a risk profile for each country (Figure 2). We found that certain land-locked countries, such as Chad (highlighted in red), would experience the highest impact of sea level rise due to their limited dependence on a few cables in neighboring nations for international connectivity. This insight is critical for designing targeted resilience measures to ensure optimized infrastructure resilience for these countries.
Evaluating Risk From Multiple Disasters
In this scenario, we used Xaminer to evaluate the effects of four disasters (earthquakes, hurricanes, sea rise, and solar storms), each with a 5% failure scenario. To achieve this, Xaminer identifies the top 5% of the locations at the highest risk of being impacted by each disaster based on its failure models and employs its workflow to determine the effects. The results from this analysis indicate that the impact on each infrastructure component is at least 15% globally. Moreover, as seen in Figure 3, regions like South America and the Balkans are at a significant risk.
Country-level Connectivity Correlation Analysis
We used Xaminer to understand if countries’ cable infrastructures are correlated based on their network layer dependence. Figure 4 shows that most clusters match regional groups or blocs. This pattern stays consistent even when countries in a cluster have different levels of physical cable infrastructure and connections to other countries or regions. These unique patterns help model general behaviors within the cross-layer cable infrastructure, which is valuable for long-term infrastructure planning and resilience analysis.
Please read our paper presented at SIGMETRICS’24 to learn more about Xaminer, its capabilities, and future developments. Xaminer codebase is open-sourced.
Alagappan Ramanathan is a PhD candidate at the University of California, Irvine, and a 2023 Pulse Research Fellow.