A common mistake when it comes to choosing a colocation provider is for enterprises to peg their decision solely on pricing metrics such as cost per rack. While not unimportant, unscrupulous new players have been known to deliberately undercut market rates, relying on the cost and difficulty of a data centre migration to substantially increase prices once the initial lease ends.
This means that a price-comparison approach without a proper consideration of the entire value proposition can result in a significantly higher TCO once the dust settles. One strategy to guard against steep price increases is to negotiate longer terms and insist on detailed rate charts ahead of signing on the dotted line, though this does not prevent price gouging on new requirements that fall outside of contracted terms.
Another more insidious consideration that can impact the TCO are requirements that were not originally envisioned by the enterprise. A prime example would be power density, which has increased dramatically with the introduction of modern hyper-converged hardware in recent years. While they can result in an energy draw of two or even three times that of traditional servers per rack, they deliver the kind of computing density and storage that is favoured for private cloud deployments.
However, an older data centre may not have the cooling capacity to support the latest generation of hyper-converged hardware. Even when colocation providers agree to them, installing them within an older facility may be a costly proposition due to substantially higher charges brought about by a low data centre power density or the need to set up specialist cooling equipment.
The cost of network connectivity is another factor that sometimes gets overlooked. Though colocation facilities typically have no shortage of connectivity, the cost can differ widely depending on the network providers with points of presence (PoP) within a facility. Some data centres may also have prebuilt fibre optic connectivity between other key facilities in the city, allowing for significantly faster provisioning times to backup data centres.