The Challenge
Operating a multi-tenant data center in Alaska presents unique connectivity challenges. With limited upstream transit options, high latency to major internet exchanges, and the critical nature of services hosted for local businesses, the network architecture had to maximize reliability, optimize performance, and maintain flexibility for growth.
My Role
I served as the principal architect responsible for the design, implementation, and ongoing operation of AS21528 (AlasConnect), a fully multi-homed autonomous system providing transit and hosting services to businesses throughout Interior Alaska.
BGP Architecture & Design
- Designed and implemented full multi-homed BGP architecture with multiple upstream providers
- Engineered traffic flows for optimal performance given Alaska’s unique connectivity landscape
- Deployed comprehensive IPv4 and IPv6 dual-stack across the entire infrastructure
- Implemented traffic engineering policies to manage cost, performance, and reliability trade-offs
Routing Operations
- Managed routing asset databases and maintained accurate IRR entries
- Coordinated with upstream ISPs and peer networks for route exchange and policy alignment
- Implemented prefix filtering and security controls to protect against route leaks and hijacks
- Developed runbooks and procedures for routing incident response
Business Operations
- Negotiated peering contracts and transit agreements
- Conducted capacity planning to ensure adequate upstream bandwidth
- Managed IP address space allocation and utilization
- Provided service assurance and SLA management for transit customers
Technical Implementation
Multi-homing Strategy
The network was architected with true multi-homing, maintaining independent BGP sessions with multiple upstream providers. This design ensured:
- No single point of failure for internet connectivity
- Automatic failover when upstream issues occurred
- Traffic optimization based on performance and cost metrics
- Flexibility to add or change providers without service impact
IPv6 Leadership
At a time when many regional providers were still IPv4-only, I led the deployment of full dual-stack IPv6 across the infrastructure. This forward-looking decision positioned the network and its customers for the evolving internet landscape.
Traffic Engineering
Given the significant cost differences between upstream providers and the performance implications of Alaska’s geography, sophisticated traffic engineering was essential. I implemented:
- Outbound traffic optimization using local preference and AS-path manipulation
- Inbound traffic control via community strings and selective announcement
- Performance-based routing for latency-sensitive traffic
- Cost optimization balancing transit costs against performance requirements
Results
AS21528 provided reliable, high-performance internet connectivity to dozens of businesses throughout the region. The architecture I designed proved resilient through numerous upstream outages and provider changes, maintaining connectivity when customers needed it most.
The multi-tenant design enabled efficient resource utilization while maintaining strict separation between customer environments, establishing a model for regional data center operations in challenging environments.