Why your colocation decision is also a connectivity decision

Ask CTOs how they evaluate a colocation provider and the conversation starts mostly the same way: location, power capacity, uptime SLAs, physical security, and so on. These are reasonable questions; they're also only half the picture. The connectivity decisions—such as how traffic moves between that facility and the rest of your infrastructure, your cloud providers, and your end users—are at least as consequential, and they tend to receive far less scrutiny than they deserve.

This imbalanced review is more inclined to backfire in APAC. The region's geography, variability of network infrastructure across markets and rapid growth of latency-sensitive workloads mean that a colocation facility that ticks every box on the checklist can still deliver a less desirable outcome if the connectivity layer has not been considered from the outset.

Designing colocation connectivity for APAC

Getting the connectivity architecture right at the colocation layer directly improves performance for the workloads that enterprise digital transformation initiatives are built around. Organisations that are building infrastructure that genuinely supports these workloads are the ones that start with the connectivity questions—what the data needs to do, where does it need to go and how quickly—and work backward to the physical infrastructure requirements from there.

Something to consider: At a colocation facility, what connectivity options are available is largely determined by which carriers and cloud providers are present inside. If your chosen facility doesn't host the carriers you want, or doesn't offer direct interconnects to your preferred cloud providers, you either pay a premium to bring that connectivity in, or have to accept an unaccounted-for performance compromise.

Beyond how the facility connects to external networks, another pertinent question to raise is how your infrastructure connects to adjacent systems within the facility itself. That's where cross connects (or direct physical connections between systems within the same facility) become relevant.

As your colocation facility connects your infrastructure to the outside world, cross connects let traffic move between your infrastructure and nearby systems, including those of cloud providers, network carriers, or business partners, without leaving the data centre. For latency-sensitive workloads—like financial transaction processing, real-time AI inference, e-commerce, content delivery, and most customer-facing applications—this is a significant factor.

For example, in financial services where transaction processing windows are measured in milliseconds, the difference between an application that performs as designed and one that doesn't is relative to whether a cross connect or a public internet path is used. The same principle applies to real-time AI inference, high-frequency data synchronisation, and any workload where the speed of data movement is a performance determinant.

In APAC, where direct cloud peering arrangements vary significantly by market, having cross connect options built into your colocation strategy from the start can steer you toward low-latency cloud access, avoiding a potentially costly workaround. However, cross connects are frequently overlooked in the initial colocation evaluation partly because they do not surface prominently during the procurement process, and partly because their impact only becomes apparent once latency-sensitive workloads are operational.

A case in point for incorporating connectivity layer early: A growing telco customer with Telstra International found that by building 100 Mbps connectivity and cross-connect services into their APAC deployment from the beginning rather than treating them as a follow-on procurement, they avoided the performance bottlenecks and budget overruns that typically come with retrofitting connectivity into an existing colocation arrangement.

The case for bundled infrastructure

A bundled infrastructure approach is the opposite of managing separate contracts with a colocation provider and a connectivity carrier. And their deciding difference? Accountability.

Picture this: When latency spikes or a link degrades, pinpointing where the issue sits occurs. This takes time away from an enterprise’s operations team to work on the issue, which end users will eventually notice.

The ambiguity of accountability is removed in a bundled approach where both the physical infrastructure and the connectivity layer is managed by the same provider. Visibility across both dimensions allows for more coherent capacity planning as the provider can model the relationship between an infrastructure footprint and an enterprise’s connectivity requirements as a single optimisation design rather than two independent ones.

Nonetheless, some enterprises stick to keeping contracts separate for the flexibility of switching carriers without changing facilities, or to negotiate connectivity terms independently. While workable, this approach demands clearly defined SLA ownership and a higher degree of internal coordination to manage issues well when they arise.

Telstra International's network spans over 400 Gbps of lit capacity across key Asia Pacific routes, with a colocation footprint of 600+ data centres in 37 countries and a 30% increase in total APAC network capacity deployed in 2025. This framework is well engineered to address gaps that surface from connectivity being treated as an afterthought rather than a design input.

Connectivity decisions under growing complexity

With enterprise infrastructure needs evolving, connectivity-first strategies are more crucial than ever. AI workloads increasingly require moving large amounts of data between edge and core, while hybrid and multi-cloud setups rely on consistent low-latency links to various cloud providers. Expansion into APAC markets with connectivity added after deployment often leads to higher costs and reduced performance. And this leads us back to why bringing your connectivity decision right up there with your colocation or any infrastructure decision will be your best bet to safeguard against complexity.