Ask an infrastructure vendor why location matters and you will hear the same word within thirty seconds: latency. Milliseconds have become the industry’s favourite unit of persuasion. Yet for most enterprise workloads, latency stopped being the deciding factor some time ago. The thing that still matters, and matters more than the industry admits, is proximity. And the two are not the same argument at all.<\/p>\n
Consider two German cities at opposite ends of the country. The network round trip between Hamburg and Munich is in the region of 8 to 10 milliseconds. For an enterprise resource planning (ERP) system, a database replication job, a virtual desktop session or a SaaS platform, that delay is imperceptible. No user will ever notice it. No SLA will ever be breached by it.<\/p>\n
Now consider the same journey by car: six to seven hours on the A7, on a good day.<\/p>\n
That asymmetry is the whole story. From a network perspective, Hamburg and Munich are next door to each other. From a human, contractual and legal perspective, they are a full working day apart. When enterprises say they want infrastructure “close by,” it is rarely the packets they are worried about. It is everything else.<\/p>\n
Let us be fair to latency first. Some workloads genuinely live and die by it: high-frequency trading, real-time industrial control, competitive gaming, certain classes of AI inference at the edge. If that is your business, every millisecond is money.<\/p>\n
But these are the exceptions. For the vast majority of enterprise applications, latency is a threshold metric. Below a certain point, typically somewhere between 10 and 30 milliseconds depending on the workload, further improvement delivers precisely nothing. The application cannot feel it. The user cannot see it. You are optimising a number on a dashboard.<\/p>\n
Proximity works differently. Its benefits do not flatten out once you cross a threshold. Being physically near your infrastructure keeps paying dividends in ways that never show up in a ping test.<\/p>\n
Serviceability and SLA credibility.<\/strong> Most managed service providers and enterprise IT teams carry contractual response commitments: an engineer on site within four hours, hardware swapped within a defined window, hands-on remediation when remote access fails. A four-hour on-site SLA is realistic when the data centre is 40 minutes from your engineering team. It is fiction when it is a flight away. Enterprises that have lived through a 2 a.m. hardware failure understand this in a way no latency chart can capture.<\/p>\n Jurisdiction and legal certainty.<\/strong> Where your data physically sits determines which courts, regulators and legal instruments apply to it. For European enterprises navigating GDPR, sector-specific regulation and growing sovereignty requirements, “close” increasingly means “inside the right legal boundary.” A data centre 5 milliseconds away in another jurisdiction can be legally further from you than one 15 milliseconds away at home. Latency measures distance in fibre. Jurisdiction measures it in law.<\/p>\n Auditability and assurance.<\/strong> Compliance teams, auditors and security officers periodically need to stand inside the facility that holds their infrastructure: to inspect physical controls, witness decommissioning, verify what the contract promises. When the site is nearby, an audit is an afternoon. When it is remote, it is a project with travel budgets, and in practice it happens less often than it should.<\/p>\n People and accountability.<\/strong> Proximity means your team knows the facility, knows the operations staff by name, and can escalate face to face when something goes wrong. It means local-language support in the local business culture. These sound like soft factors until an incident makes them very hard ones.<\/p>\n Data gravity and migration reality.<\/strong> Moving workloads is easier when the physical logistics are simple. Racking equipment, staging migrations and running hybrid setups all get cheaper and faster when the facility is within driving distance of your own sites.<\/p>\n The hyperscale era trained the market to accept a trade: consolidate everything into a handful of massive, distant campuses, and let fast networks paper over the distance. For pure performance, the trade often works. The network really is that good.<\/p>\n But the trade quietly assumed that performance was the only thing distance affected. It is not. Every kilometre between an enterprise and its infrastructure adds cost and friction to servicing, auditing, staffing, legal assurance and operational trust. None of that friction appears in a latency figure, which is exactly why the industry prefers to talk about latency.<\/p>\n This is, ultimately, an argument for a different mental model of edge infrastructure. The edge is usually sold as the place where latency goes to die. The better definition: the edge is where infrastructure comes back within reach of the people, contracts and laws that depend on it.<\/p>\n An enterprise in Munich does not need its data centre to shave 6 milliseconds off a round trip to Hamburg. It needs an engineer who can be on site before lunch, an auditor who can visit without booking flights, a contract governed by the jurisdiction it already operates in, and an MSP whose SLA is grounded in geography rather than optimism.<\/p>\n Latency asks how fast your data can travel. Proximity asks how fast your people, your lawyers and your engineers can. For most enterprises, the second question is the one that decides where infrastructure belongs.<\/p>\n nLighten operates more than 30 edge data centres across Germany, France, the UK, the Netherlands, Belgium, Switzerland and Spain: infrastructure close to where European enterprises actually work.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":" The Hamburg to Munich test: 8 milliseconds by fibre, 7 hours by car. Why physical proximity, not latency, decides where enterprise infrastructure belongs.<\/p>\n","protected":false},"author":32,"featured_media":30077,"template":"","categories":[],"tags":[],"class_list":["post-30066","blog","type-blog","status-publish","has-post-thumbnail","hentry"],"acf":[],"yoast_head":"\nThe uncomfortable question for centralised models<\/h2>\n
Redefining “edge” as nearness to people, not just packets<\/h2>\n