How would an Accenture network engineer design a highly available and resilient network architecture for a large enterprise?

Designing a highly available and resilient enterprise network requires a layered approach, prioritizing redundancy at every critical point and implementing intelligent routing protocols. The goal is to eliminate single points of failure, ensure fast failover, and maintain optimal performance under various load and fault conditions. This typically involves a hierarchical design like the three-tier model (core, distribution, access) or a spine-and-leaf architecture for data centers, coupled with robust hardware and software redundancy strategies.

Key Design Principles

Achieving high availability involves several principles. Hardware redundancy is crucial, encompassing dual power supplies, redundant modules, and active-standby or active-active configurations for critical devices like routers and firewalls. Link redundancy ensures multiple physical paths between network devices, often using technologies like EtherChannel or Link Aggregation Groups (LAGs). Protocol-level redundancy, such as HSRP, VRRP, or GLBP, provides gateway fault tolerance. Routing protocol choices, like OSPF or EIGRP for interior gateways and BGP for external connectivity, must support rapid convergence and path diversity. Finally, network segmentation using VLANs and careful IP addressing planning helps contain broadcast domains and improve security.

Best practice

Implement active-active redundancy wherever possible for both devices and links. This not only provides fault tolerance but also utilizes both paths, enhancing bandwidth and reducing latency compared to active-standby configurations. Employ dynamic routing protocols configured for fast convergence, such as BFD (Bidirectional Forwarding Detection) with OSPF, to minimize detection and failover times to under a second. Regularly test failover scenarios through planned maintenance windows to validate the resilience design.

Edge case interviewers probe for

Interviewers often ask about handling “split-brain” scenarios in active-active setups, where both redundant devices believe they are the primary, leading to network instability. A strong answer would explain using robust heartbeating mechanisms, dedicated out-of-band management networks, and proper tie-breaking logic (e.g., higher priority, preemption delays, or specific failure detection protocols) to prevent such situations and ensure graceful failover. Another edge case is designing for partial path failures that might still impact service without full link loss.

Common mistake

A common mistake is focusing solely on hardware redundancy without considering the software and protocol layers. For instance, having two firewalls in active-standby is good, but if their state synchronization fails or routing protocols do not converge quickly enough, there will still be a significant outage during failover. Another error is neglecting comprehensive network monitoring, which is essential to detect subtle performance degradations or incipient failures before they become critical.

What the interviewer is checking

The interviewer is assessing your understanding of fundamental networking concepts, your ability to design robust and scalable solutions, and your practical experience with redundancy mechanisms. They want to see if you can think beyond basic connectivity to address fault tolerance, performance, and operational aspects like monitoring and troubleshooting in a large-scale environment. Your structured approach to problem-solving and awareness of potential pitfalls are key.