Capacity planning across 2,000+ collocations

The network expanded nationwide while service held steady and operations stayed repeatable.

Cloud & infrastructure / Reliability / Security

Problem

A broadband network spread over 2,000+ collocations had to keep adding customers and services, including VoIP, without degrading the service it already carried.

Constraints

Nationwide footprint, long hardware lead times
Several network generations running side by side (ATM, IP, VoIP)
Security services running on shared infrastructure

Architecture & approach

Plan capacity from measured utilization rather than forecasts alone. Standardize integration and deployment steps so expansions repeat cleanly, and build the process, training and escalation paths that let operations scale beyond a few experts.

Fig. 306-2Carrier network hierarchy Critical path

Key decisions

How should capacity be added? 306-3.1

RejectedReact to congestionCustomers find the problem before you do.
ChosenPlan from measured utilizationOrder ahead of lead times, where the data says growth is.

Outcome

Integration and expansion across a 2,000+ collocation footprint
Process, training and escalation for network security and IP services

Technology

ATM, IP, VoIP, OC transport, Network security services, Capacity planning

Lessons & tradeoffs

Reliability is mostly process. Heroics do not scale to 2,000 sites.
Everything I know about systems thinking started here.