Halfway through a cloud migration, or when a product launch is being coordinated across multiple time zones, a connectivity outage isn’t an inconvenience. It’s a financial event. And teams that treat internet access as a commodity to be sourced at the lowest price tend to discover this at the worst possible moment.
Start by finding your single points of failure
Before a high-stakes project starts, you need someone to make that warts-and-all audit of your network infrastructure. Not the box-ticking compliance audit – the one where you explain what the consequence is if you cut a specific cable, or if a switch goes, or if a provider exchange is offline.
The vast majority of businesses on run-of-the-mill broadband have one single point of failure at least that they don’t know about. If it’s not the last mile, statistically the most vulnerable point of any network path, it’s the router with no failover. Or the ducting that the primary and secondary link share, knowing that if there’s one good dig that’s both your connections down.
Map it before the project starts. Any node where one failure cascades into full outage, you sign off the project with a mitigation strategy attached to it.
Choosing the right infrastructure partner
Choosing a supplier in this context is a strategic decision, not a procurement one. The opportunity cost of IT downtime averages $5,600 per minute (Gartner), and escalates significantly for large-scale environments. An SLA offering 99.9% uptime might leave you thinking ‘what does “best effort” really mean to my business’?
The contrast between enterprise- and residential-grade support is more than just a piece of paper. A business internet service provider with a focus on business connectivity will typically provide you with a contractually guaranteed SLA including a committed Mean Time to Recovery of four hours or fewer, and give you direct access to engineering staff rather than a generic customer support hotline. This is an entirely different way of doing business to an ISP whose largest proportion of customers are end-users in their homes.
When vetting suppliers for high-impact projects, check whether they can guarantee MTTRs, find out whether support tickets will be handled urgently by a designated stakeholder, and whether those support desks are answering the phones in the provider’s office or in a facility several time zones away.
It’s also worth asking what their proactive monitoring looks like. Some providers will flag a degrading connection before you’ve even noticed a problem. Others wait for you to ring them.
Dedicated access over shared bandwidth
Want to discuss an issue that no one ever seems to think about until it becomes a showstopper? Bandwidth contention. Shared broadband sounds innocuous enough but remember, it means that your uploads are competing with every other business on the block. You won’t notice it most of the time but, sure as fate, you’ll feel the pinch when you’re sending that big bit of code, or streaming that global team meeting in HD.
Dedicated internet access removes that worry. Your connection is uncontended; the ‘pipe’ is all yours, no matter what your neighbours are up to. Should you ever be in any doubt as to what your nearest and dearest are doing with their bandwidth, it’s nice to know that only their activity holds any surprises for you.
This also plays into latency. Losing the odd packet, or seeing one or two latency spikes a day when using YouTube is probably something you can live with. The same can’t be said of a VoIP-heavy project conference call or a real-time deployment. A decent, enterprise-grade connection is designed to keep both to a minimum. Regular broadband? Not so much.
Symmetrical upload and download speeds matter here too. Standard broadband gives you upload speeds that are a fraction of the download, which falls apart when a team is pushing large files to the cloud or running concurrent outbound video feeds.
Build a diverse carrier strategy
Having one solid SLA connection is better than having none. But one additional connection sharing the same duct, exchange, and cables is almost no improvement at all.
A proper diverse carrier strategy means sourcing a secondary link that doesn’t share the same duct, exchange, or network topology as the primary. That might mean combining a fiber leased line with a 5G failover, or using two providers whose routes into the building come from different physical directions.
Failover systems can automate the switch between primary and secondary connections so that the cutover is invisible to users. When it’s configured correctly, most team members won’t notice the primary went down. When it isn’t configured at all, everyone notices at once.
This isn’t overengineering. It’s the same logic as having a backup generator – you don’t use it every day, but the day you need it, you really need it.
The war room protocol
It is a good practice to have a pre-prepared escalation document for every high-stakes project, which is basically a short list of who to call, in what order, with direct numbers for technical contacts at each provider – not the general helpdesk number.
The account manager’s direct line, the on-call engineering contact, and any out-of-hours emergency paths, etc. Should be included in this document to make the critical window calling easy and fast.
This sounds basic. Most teams don’t have it. When something fails during a critical window, the time spent searching for contact details or navigating phone trees is time that extends the outage.
A Disaster Recovery Plan for connectivity doesn’t need to be a long document. It needs to be specific, pre-verified, and accessible to more than one person on the team. Print a copy as well — if the network’s down, a document saved on a shared drive isn’t much help to anyone.
Business connectivity decisions made months before a project begins will either protect the work or expose it. The infrastructure either holds under pressure or it doesn’t – and finding out which during the project itself is not a position anyone wants to be in.
