[Paper Review] Can SDN Mitigate Disasters?

Vincent Gramoli, Guillaume Jourjon+1 more|arXiv (Cornell University)|Oct 16, 2014

Cloud Computing and Resource Management7 references3 citations

TL;DR

This paper proposes Software Defined Edge (SDE), an SDN-based approach deployed at the network edge to achieve zero recovery point objective (RPO) in disaster recovery for geo-replicated datacenters. By leveraging application-level information and centralized control, SDE enables a fault-tolerant key-value store to recover from disasters in under 30 seconds, demonstrating that SDN-level failure detection significantly enhances application resilience.

ABSTRACT

Datacenter networks and services are at risk in the face of disasters. Existing fault-tolerant storage services cannot even achieve a nil recovery point objective (RPO) as client-generated data may get lost before the termination of their migration across geo-replicated datacenters. SDN has proved instrumental in exploiting application-level information to optimise the routing of information. In this paper, we propose Software Defined Edge (SDE) or the implementation of SDN at the network edge to achieve nil RPO. We illustrate our proposal with a fault-tolerant key-value store that experimentally recovers from disaster within 30s. Although SDE is inherently fault-tolerant and scalable, its deployment raises new challenges on the partnership between ISPs and CDN providers. We conclude that failure detection information at the SDN-level can effectively benefit applications to recover from disaster.

Motivation & Objective

• To address the limitation of existing fault-tolerant storage systems that cannot achieve zero recovery point objective (RPO) due to data loss during migration.
• To explore whether SDN can be leveraged at the network edge to improve disaster resilience in distributed systems.
• To design and evaluate a fault-tolerant key-value store that achieves near-instantaneous recovery using SDE.
• To investigate the operational challenges in deploying SDE, particularly the collaboration between ISPs and CDN providers.
• To demonstrate that failure detection at the SDN control plane can effectively support application-level disaster recovery.

Proposed method

• Propose Software Defined Edge (SDE), an SDN architecture deployed at the network edge to manage and optimize data replication and failover.
• Integrate application-level semantics into the SDN controller to guide intelligent routing and replication decisions.
• Implement a fault-tolerant key-value store that leverages SDE for real-time data synchronization across geo-distributed datacenters.
• Use centralized SDN control to detect network failures and trigger immediate failover and data recovery procedures.
• Design the system to minimize data loss by ensuring data is consistently replicated before migration is initiated.
• Evaluate the system in a real-world testbed to measure recovery time and resilience under disaster conditions.

Experimental results

Research questions

• RQ1 Can SDN at the network edge eliminate data loss during disaster recovery in geo-replicated datacenters?
• RQ2 How effectively can SDE reduce recovery time compared to traditional replication mechanisms?
• RQ3 What are the operational challenges in deploying SDE, particularly in the context of ISP and CDN collaboration?
• RQ4 Can failure detection at the SDN level be used to trigger application-level recovery with zero RPO?
• RQ5 What performance and reliability guarantees does SDE provide under real-world disaster scenarios?

Key findings

• The proposed SDE-based key-value store achieved disaster recovery in less than 30 seconds, demonstrating near-instantaneous failover.
• SDE enables a zero recovery point objective (RPO) by ensuring data is consistently synchronized before migration.
• Failure detection at the SDN control plane significantly improves the responsiveness and accuracy of disaster recovery mechanisms.
• The system is inherently fault-tolerant and horizontally scalable, making it suitable for large-scale datacenter deployments.
• Deployment of SDE introduces new challenges in coordination between ISPs and CDN providers, particularly around trust and data ownership.
• The integration of application-level semantics into SDN control logic enables more efficient and resilient data replication strategies.