Most organisations believe their connectivity problem is solved. Carrier contracts are in place, SD-WAN is deployed, and redundancy boxes have been ticked. Yet when a critical application degrades or a regional outage halts operations, a different reality emerges: carriage is not background plumbing. It is a performance, resilience and risk multiplier – and it is still widely managed as if it isn’t.
Cloud-first operating models have fundamentally changed what depends on connectivity decisions. SaaS platforms, real-time collaboration, security inspection services and emerging AI workloads all assume consistent, predictable network performance. But connectivity is often still procured using decade-old logic – optimised for price, incumbency and contract simplicity – with limited visibility into how those decisions constrain everything layered on top.
The misconception that quietly drives outages
Carriage is frequently treated as a commodity: interchangeable, largely invisible, and won on price. This perception persists because connectivity rarely announces itself as the root cause. When Teams calls degrade or Salesforce feels sluggish, the application or cloud provider is blamed. When outages occur, attention shifts to data centres or hyperscalers.
Industry data tells a different story. The Uptime Institute Annual Outage Analysis 2024 identifies networking and connectivity issues as one of the most common contributors to end-to-end IT service disruptions, accounting for around 30 per cent of reported outages (Uptime Institute, 2024).
Performance degradation is even harder to diagnose. Latency, jitter and packet loss directly impact SaaS responsiveness, video quality and security inspection effectiveness – regardless of how well the underlying cloud platform is performing (ThousandEyes, n.d.; Kentik, n.d.). Even very low levels of packet loss can materially reduce usable throughput once encryption and security inspection overhead is introduced, particularly in cloud-delivered security architectures (Cloudbrink, 2024). Yet most organisations lack the end-to-end visibility required to connect these symptoms back to carriage design decisions.
Why carriage now determines outcomes
This dependency was not always so pronounced. When applications lived in on-premises data centres and traffic flows were predictable, connectivity requirements were simpler. Today’s operating environment is materially different:
- SaaS dependency is total. Email, CRM, finance and collaboration platforms now sit outside the enterprise perimeter and are accessed over networks the organisation does not control.
- Real-time interaction is non-negotiable. Video conferencing, cloud contact centres and remote desktops demand consistent low latency; degradation is immediately visible to users and customers (ThousandEyes, n.d.).
- Security inspection adds unavoidable overhead. Secure access services, inspection and monitoring introduce additional latency on every packet, amplifying the impact of poor underlying carriage.
- AI workloads increase sensitivity. Data movement, inference and real-time analytics require high-throughput, low-jitter connectivity that many legacy designs were never intended to support.
In this context, carriage decisions set the ceiling for performance and resilience. No amount of SD-WAN optimisation or application tuning can compensate for inadequate path diversity, chronic congestion or sub-optimal routing.
Evidence snapshot: what the data shows
Networking and connectivity issues account for approximately 30% of IT service outages, making them one of the most common contributors to service disruption (Uptime Institute, 2024).
Latency and packet loss directly degrade SaaS performance, collaboration quality and security inspection effectiveness, independent of cloud provider availability (ThousandEyes, n.d.; Kentik, n.d.).
Single-carrier designs and shared physical routes materially increase correlated failure risk, expanding outage blast radius during provider faults (Uptime Institute, 2022; Lightyear, n.d.).
Most organisations lack end-to-end visibility linking application experience to specific network paths, leading to carriage issues being misdiagnosed as application or cloud failures (industry monitoring research).
How poor carriage design amplifies failure
When carriage is treated as a commodity, failure tends to manifest in three ways.
Latency becomes user experience. Congested or inefficient paths make applications feel slow and unreliable, even when headline bandwidth appears sufficient. Users experience this as application failure, but the cause often lies in routing decisions made years earlier.
Outages expand their blast radius. Single-carrier designs – or “redundant” circuits sharing the same physical infrastructure — create correlated failure risk. When that carrier experiences a fault, both paths can fail simultaneously. By contrast, genuinely diverse, multi-carrier designs with separated physical routes materially reduce outage impact and recovery time (Uptime Institute, 2022).
Congestion erodes security visibility. When carriage capacity is constrained, security inspection platforms may drop packets or bypass analysis to preserve throughput. Connectivity appears intact, but threat visibility quietly degrades — creating risk without obvious failure signals (Kentik, n.d.).
What resilient carriage actually requires
Resilience is not achieved through redundancy alone. It requires diversity, visibility and architectural alignment.
Diversity means independent failure domains. Two circuits from the same carrier – or different carriers sharing the same physical route – provide a false sense of protection. Resilient design demands genuinely independent paths: different carriers, separate physical infrastructure and, where appropriate, different access technologies. In the Australian context, this may include combinations of fibre, fixed-line and cellular connectivity with clearly separated ingress points (ACCC, n.d.).
Visibility must extend end to end. Organisations need the ability to correlate application performance with specific network paths. Can spikes in collaboration issues be traced to congestion on a particular carrier? Do teams understand which SaaS services traverse which links? Without this visibility, troubleshooting becomes reactive and imprecise (ThousandEyes, n.d.).
Design must align with SD-WAN and cloud architecture. SD-WAN platforms can only make intelligent decisions if meaningful path diversity exists. Similarly, private connectivity services to cloud providers deliver resilience benefits only when the underlying carriage is independently designed.
Many performance issues attributed to SaaS platforms are, in reality, carriage problems in disguise. Any serious conversation about application resilience must start with the layer underneath.
What leaders should reassess now
For executives accountable for uptime and operational performance, carriage warrants renewed scrutiny:
- Validate diversity – don’t assume it.
- Instrument for insight.
- Revisit legacy procurement decisions.
Carriage rarely attracts headlines, but it determines whether everything else performs when it matters. Redundancy without diversity is not resilience – it is false confidence. And in an environment where network-related issues drive a significant proportion of service disruptions, that confidence carries a measurable cost.
If this article has raised questions about how your own carriage environment is supporting performance or resilience, reach out to one of our experts for a confidential discussion.
Sources & further reading
- Uptime Institute — Annual Outage Analysis 2024
- Uptime Institute — 2022 outage analysis press release (includes networking as a major contributor to IT service downtime)
- ThousandEyes — SaaS monitoring
- ThousandEyes — Enterprise digital experience
- Kentik — Network latency (Kentipedia)
- Lightyear — Redundancy vs diversity guide
- Cloudbrink — Packet loss and network performance (29 Feb 2024)
