When a Router Refresh Goes Wrong: A Telecom's Multi-Vendor Decision
Archi3 min read
Contents
The choice that shaped a network's fate#
In late 2024, a mid-sized internet service provider (ISP) serving about 40,000 homes faced a hardware refresh. Its core routers, installed in 2017, were approaching end-of-support. The network team narrowed the options to two vendors: one offered a fully integrated stack with a single management interface, the other a modular system that mixed and matched components from different suppliers. The decision would affect how quickly the ISP could recover from outages, how much its engineers needed to learn, and how much it would pay over five years.
This is a plain-language look at how that ISP weighed the trade-offs, based on interviews with network engineers and public documents. No single answer fits every company, but the reasoning shows what to ask before signing a multi-year contract.
The appeal of a single vendor#
The integrated option promised simplicity. One vendor would supply hardware, software, and support. The ISP's engineers would learn one command-line interface and one set of troubleshooting tools. The sales pitch emphasized faster incident resolution because there would be no finger-pointing between suppliers when something broke.
The catch was lock-in. Upgrades would follow the vendor's schedule, and the ISP would have little room to negotiate on price after the initial term. The integrated system also carried a higher upfront cost: about 20 percent more than the modular option, according to the ISP's internal estimate.
The modular alternative#
The modular approach let the ISP keep its existing switches and firewalls while replacing only the routing layer. It could buy optical transceivers from a third party and use open-source monitoring tools. The network team liked the flexibility and the lower initial price. But modularity brought complexity: more interfaces to test, more firmware versions to track, and more potential for compatibility problems.
The ISP's lead engineer recalled a previous incident where a firmware update on one vendor's line card clashed with another vendor's switch, causing a brief but widespread outage. That memory weighed on the decision.
How they decided#
The team ran a two-week pilot with each option. They simulated a fiber cut and measured how long it took to reroute traffic. The integrated system recovered in 45 seconds; the modular system took 90 seconds because of extra steps to synchronize routing tables across vendors. But the modular system cost 15 percent less per year in licensing and allowed the ISP to reuse existing hardware, saving about $200,000 upfront.
In the end, the ISP chose a hybrid: integrated routing for the core, modular for the edge. The core needed predictable recovery times and single-vendor support. The edge could tolerate slightly longer failover and benefited from cost savings.
The decision was not about picking the "best" technology. It was about matching risk tolerance to each part of the network. The core carried the most traffic and the highest cost of downtime, so the ISP paid for simplicity. The edge was more distributed and easier to troubleshoot, so flexibility won.
What other companies can learn#
Map your failure modes. Before comparing vendors, list the outages you have experienced and how long they lasted. That tells you where recovery speed matters most.
Price the exit, not just the entry. Ask vendors for the cost of migrating away after three years. Lock-in often hides in support contracts and proprietary interfaces.
Test with your own traffic. Pilot both options under real load. Lab tests miss the quirks of production traffic.
Check the support model. A single vendor may resolve issues faster, but only if its support is competent. Ask for references from similar-sized customers.
Plan for skills. Modular systems require broader expertise. If your team is small, the integrated option may reduce training and hiring costs.
The ISP's choice was not permanent. It scheduled a review after two years to see if the hybrid still made sense. Network needs change, and so do vendor offerings. The key is to make the decision with clear eyes about trade-offs, not marketing promises.
More articles