Infrastructure
Your Internet Connection Has a Single Point of Failure. Here's How to Fix That.
Most businesses know who their internet provider is but not what happens when that provider has an outage. This guide covers shared vs. dedicated internet, what failover actually buys you, what SD-WAN does differently from MPLS, and a short checklist to run against your current setup.
Most businesses can tell you who their internet provider is. Far fewer can tell you what happens to the business the moment that provider has an outage. If the answer is "everyone stops working until it comes back," the network has a single point of failure sitting in a closet somewhere, and it is worth thirty minutes to understand the options for fixing that before an outage forces the question.
This guide covers the practical differences between shared and dedicated internet, what failover actually buys you, what SD-WAN is and is not, and a short checklist to run against your current setup.
Shared Business Internet vs. Dedicated Internet Access
Most business internet circuits, most cable plans and a lot of fiber plans, are shared or "best-effort" connections. You get a bandwidth ceiling, but real-world throughput varies with local demand, and the provider makes no contractual promise about uptime, latency, or packet loss. It is usually the least expensive option, and for a lot of small offices it is genuinely good enough.
Dedicated Internet Access (DIA) is a different product, not just a faster version of the same one. It is dedicated, committed bandwidth with symmetric upload and download speeds, not shared with other subscribers the way a best-effort circuit is, backed by a service level agreement that commits the carrier to specific uptime, latency, and packet-loss targets, with financial remedies when the carrier misses them. Your traffic still travels the public internet once it leaves the carrier's network, and the physical plant may still be shared with other customers; what DIA guarantees is your access bandwidth and the carrier's contractual performance commitment, not a physically private circuit. That contractual backing is the actual value. It costs more, so it makes the most sense for sites hosting services other people depend on, sites with compliance or insurance documentation requirements, and any location where a slow or flaky connection has a direct, immediate cost.
Neither option is automatically wrong for a given business. The mistake is not knowing which one you have, and not deciding on purpose.
What Failover Actually Buys You
Failover means a second, independent path to the internet that takes over automatically when the primary circuit degrades or fails. Two versions of this are common:
A second wired circuit, often a lower-cost cable connection backing up a dedicated fiber primary, gives you a second internet source but shares physical infrastructure risk with the first (many wired outages, from a cut cable to a provider-wide event, can affect more than one circuit from the same carrier or the same local plant).
LTE or 5G cellular failover uses a different access medium, the cellular network instead of your wired circuit, which is why it is the more common choice for mission-critical sites: a backhoe that cuts one buried cable is far less likely to also take down a cell tower. That said, the tower itself typically still depends on its own wired backhaul and the local power grid, so cellular failover is access-medium diversity, not a guarantee of total independence. For a genuinely mission-critical site, it is worth confirming the cellular provider's backhaul and power setup rather than assuming "cellular" alone means fully separate infrastructure.
Failover only works if it is actually configured to switch over automatically and if someone is monitoring whether it is doing its job. A failover device sitting unconfigured in a closet, or one nobody has tested since installation, provides the appearance of redundancy without the substance.
SD-WAN, and How It Differs From MPLS
SD-WAN (Software-Defined Wide Area Network) is a technology layer that sits above your internet circuits and makes routing decisions based on circuit health and application requirements. Configured for active-active use, it can route voice and video down the cleanest circuit while less time-sensitive traffic takes another, and rebalance quickly if one degrades. That is a configuration choice, not something every SD-WAN deployment does automatically: some are set up active-standby, or with policies that pin certain traffic to a preferred link, so a second circuit sits idle until needed. For a business running two or more circuits at a site, the difference between those configurations is exactly what is worth asking about before assuming a second connection is being put to work.
MPLS is the older approach to a related problem: a carrier-managed private network that keeps traffic isolated from the public internet, at a materially higher price and with slower provisioning for new sites. The quality-of-service guarantees usually associated with MPLS come from the carrier's traffic-class configuration and the specific contract, not from label switching itself, so it is worth confirming what QoS commitments an MPLS offering actually includes before assuming they come standard. SD-WAN and MPLS are not strictly either/or: SD-WAN is an orchestration layer that can run over MPLS circuits, ordinary internet circuits, or a mix of both. What most organizations mean by "moving off MPLS" is replacing expensive private MPLS transport with SD-WAN-orchestrated internet circuits instead, with traffic routed directly to cloud services rather than backhauled through a central hub first. How much resilience that delivers depends on the number and diversity of those circuits, their individual SLAs, and the failover policy configured, not on adopting SD-WAN by itself. Regulated environments that need carrier-enforced traffic isolation sometimes keep MPLS transport for that reason, with or without SD-WAN orchestrating it.
Neither one is a default correct answer. The right choice depends on how many sites you run, what applications matter most, and whether a regulatory or contractual requirement specifically calls for traffic isolation.
Reliability, Voice, and Video
Four measurements determine whether a connection actually feels reliable, and none of them is simply "is it up or down." Latency is the round-trip delay. Jitter is how much that delay varies from one moment to the next. Packet loss is the percentage of data that never arrives. Utilization is how much of the circuit's capacity is actually in use; a circuit that is undersized or saturated during peak hours can produce the same choppy calls and frozen video as a circuit with healthy latency, jitter, and packet loss but simply not enough headroom. Real-time applications like VoIP calls and video conferencing degrade noticeably once any of these crosses a threshold, often well before a circuit fully fails and shows up as an obvious outage. A connection can be technically "up" and still be the reason calls are choppy and video meetings freeze.
A Short Checklist for Your Current Setup
- Confirm whether your primary circuit is shared/best-effort or Dedicated Internet Access, and whether that matches what the business actually needs.
- Find out whether you have a second path to the internet, and whether it is wired failover sharing infrastructure with the primary or a different access medium like cellular.
- Ask when failover was last tested with a real cutover, including a controlled upstream or health-check failure and not only a disconnected cable, since a link stays electrically up during a carrier or routing outage even though it is no longer passing traffic.
- Check whether anyone is actively monitoring latency, jitter, packet loss, and utilization on your circuits, or only finding out about problems when someone complains.
- If you run more than one site, or more than one circuit at any site, find out whether traffic is actively balanced across them (SD-WAN or similar) or whether the second connection just sits idle until the first one fails.
If you cannot answer one or more of these with confidence, that is the gap worth closing first.
Where Cyber One Solutions Fits
We source and manage internet circuits from a broad range of national and regional carriers, matched to what actually serves each site, so you have a single point of contact instead of a separate relationship and escalation path for each provider. Our managed internet services cover circuit sourcing and SLA negotiation, LTE and 5G failover deployment, SD-WAN design and management, and continuous monitoring of latency, jitter, packet loss, and utilization, with our team detecting an outage within minutes and opening and managing the carrier trouble ticket once it is confirmed, so you are not the one on hold with the carrier. For a broader look at how to evaluate a networking provider, our Networking Buyers Guide covers the questions worth asking before you sign a contract.
If you are not sure whether your current setup has a real failover path or just a device that has never been tested, contact us and we will take a look.
Frequently Asked Questions
Is Dedicated Internet Access worth the extra cost for a small office?
It depends on what the office actually does. If a slow connection is an inconvenience, shared business internet is often genuinely sufficient. If the office hosts services other people depend on, has compliance or insurance documentation needs, or simply cannot tolerate inconsistent performance during business hours, the SLA backing behind DIA is what you are paying for, not just faster numbers on paper.
Do we need SD-WAN if we only have one office?
Usually not in the full multi-site sense. SD-WAN's biggest advantage, intelligently routing traffic across multiple sites and circuits, matters most once you have more than one location or more than one circuit to balance. A single office with a single circuit benefits more from getting failover right than from SD-WAN specifically.
How is cellular failover different from just having two internet providers?
The difference is access-medium diversity, not full independence. Two wired circuits from different providers can still share risk if they run through the same conduit, the same local plant, or are affected by the same regional event. Cellular failover uses a different access medium than a wired circuit, which is why it is the more common choice when any downtime at a site is genuinely unacceptable, though the tower itself still typically depends on its own wired backhaul and power, so confirming a provider's backhaul and power diversity still matters for a site where that failure mode is unacceptable.
What is the fastest way to tell if our current failover actually works?
Ask when it was last tested with a real cutover, not just reviewed on paper or confirmed present in a rack. A failover device that has never been switched over live is an unknown, not a safeguard, until someone actually forces the switchover and confirms the backup takes over automatically. Testing this by disconnecting the primary cable only proves link-down failover works; it does not confirm the health checks that should also trigger failover during an ISP or upstream routing outage, where the link often stays electrically up even though it has stopped passing traffic. A thorough test includes a controlled upstream or health-check failure, not only a pulled cable.