Essential knowledge about modern VPN technologies

Essential knowledge about modern VPN technologies

When evaluating remote connectivity solutions, understanding the difference between a robust corporate IPsec implementation and a consumer-grade VPN is paramount: while both offer encryption, I strongly recommend relying on dedicated IPsec services whenever compliance or data integrity are non-negotiable. The typical commercial VPN service works adequately for masking your general location—a functional convenience. However, when an organization requires a secure communications mechanism built directly into the network layer to protect specific internal resources, the structure and controls inherent in IPsec prove far more reliable.

The fundamental difference between a VPN and a simple proxy connection is that one encrypts the session, while the other merely masks the destination.

A common misunderstanding is confusing a Virtual Private Network (VPN) with a proxy server; they are not interchangeable tools. A proxy generally operates at an application layer—it acts like a middleman for specific types of traffic (like HTTP requests), changing only your apparent IP address for that single application stream. It provides little to no guarantee regarding the integrity or encryption of the data payload itself, making it vulnerable if the connection is intercepted outside its scope.

In contrast, a VPN is defined as a virtual network built on top of existing physical networks that can provide a secure communications mechanism for all data and information transmitted. This typically involves encapsulating your entire traffic stream—not just web requests—into an encrypted tunnel. The key difference lies in scope: the proxy manages where the request goes; the VPN secures the entire path. If you are dealing with sensitive proprietary data, or if your organization is mandated to maintain a specific level of security that might exceed what a basic layer-seven application proxy can enforce, the full encryption capability of a VPN becomes mandatory.

The limitation here is understanding that while some consumer services use protocols like OpenVPN or WireGuard which are superior to traditional proxies, they still rely on the underlying operating system's network stack. If the endpoint device itself is compromised—say, by malware—the strongest tunnel in the world can be bypassed at the source, regardless of how many points an antivirus product scored; even achieving the minimum overall score out of 18 points required for a Windows 11 antivirus product to earn AV-TEST’s "TOP PRODUCT" designation does not protect against endpoint compromise.

IPsec remains the most foundational and widely used method for building secure network tunnels.

When speaking about robust, enterprise-grade connectivity, Internet Protocol Security (IPsec) is the backbone technology you are dealing with. IPsec isn't just another feature; it is a core suite of protocols that operate at the network layer—Layer 3—of the OSI model. This placement means that IPsec can encrypt and secure all traffic passing through the tunnel, regardless of the application generating it, which is why NIST states that the most common use of IPsec implementations is providing VPN services.

An IPsec VPN establishes a highly controlled connection between two points—often an individual remote device (the client) and the organization’s designated gateway (the IPsec gateway). This gateway acts as the secure endpoint, receiving all traffic sent through the tunnel. The process involves negotiating security parameters, authenticating both ends to ensure they are who they claim to be, and then establishing cryptographic keys for symmetric encryption. This structure ensures that even if the public internet path is monitored by bad actors, the data payload remains indecipherable.

However, because of its power and complexity, IPsec requires careful configuration. The inherent difficulty lies in managing the interplay between various protocols (like IKEv1 vs. IKEv2) and ensuring proper firewall rule sets on the gateway side. Misconfiguration can lead to severe vulnerabilities; for instance, if internal resources are accessed without correctly validating the user’s machine certificate or IP address range, an attacker could potentially exploit a lateral movement vector that bypasses the tunnel entirely.

Split tunneling drastically changes traffic flow, often compromising security for perceived efficiency gains.

Split tunneling is perhaps the most controversial feature in modern VPN design because it introduces a functional trade-off between convenience and absolute security. By default, many remote access solutions are designed to route all corporate traffic through the secure tunnel. Split tunneling modifies this behavior by allowing an organization-specific subset of traffic (e.g., accessing internal servers) to flow through the VPN tunnel while simultaneously permitting other unrelated traffic (like browsing a personal news site or accessing cloud storage) to use a separate, non-VPN connection.

The appeal of split tunneling is clear: it reduces load and improves efficiency for the end user by preventing every single byte of data—from checking email to streaming video—from being forced through the secure corporate pipe. However, NIST strongly advises against this practice for IPsec VPN clients because it introduces serious security complications and risks. The common belief that "Split tunneling is always safer because less traffic goes through the VPN" is demonstrably false; in fact, the guide warns that split tunneling can expose traffic outside the VPN boundary.

When an organization implements this, they are effectively telling their network that only internal-resource traffic requires protection from the tunnel. Anything else—the non-VPN path—is exposed to the inherent risks of the public internet. While it provides functionality, its use severely limits an organization’s ability to inspect and protect all data streams passing through the connection.

A VPN fundamentally creates a virtual layer over existing physical networks.

At its core, a VPN (which stands for Virtual Private Network) is not about creating secret lines; it is about creating an illusion of privacy and security on top of existing, insecure infrastructure—the public internet. It functions by establishing encrypted tunnels that make the connection behave as if the remote user was physically plugged into the secure corporate LAN, even though they are sitting in a coffee shop.

The mechanism is straightforward: all data packets leaving your device destined for the private network are intercepted and then wrapped (encapsulated) inside an encrypted layer. This process ensures that even if someone intercepts the physical signal—say, by monitoring Wi-Fi traffic—all they see is a stream of unintelligible, scrambled data. The destination endpoint on the other side of the tunnel is responsible for decrypting that packet and forwarding it to its correct internal recipient.

The crucial conceptual shift here is moving from thinking about "secure internet access" to thinking about network encapsulation. When you use a VPN, you are essentially telling your operating system's network stack: "Do not send this data normally; instead, process it through the IPsec tunnel endpoint first." This mechanism provides the secure communications necessary for both nodes on the same physical network or two different networks entirely. The trade-off is complexity—the client and the gateway must be perfectly synchronized regarding encryption algorithms and key management to ensure a continuous, stable connection.

Utilizing a dedicated VPN router provides an essential, centralized point of security enforcement.

While modern laptops can run consumer-grade VPN software that handles tunneling at the endpoint level, for small or medium businesses (SMBs), implementing connectivity through a dedicated hardware appliance—a specialized VPN router or firewall—is vastly superior for maintaining consistent, enforceable security policies. This device becomes the network's single point of entry and exit for all remote traffic.

When you route VPN services through such an appliance, you gain centralized control over authentication, encryption protocols, and access rules that are simply unavailable when relying on individual software clients. Instead of managing client settings across multiple machines—each with its own potential configuration drift or update delay—the router enforces the policy at the gateway level. It ensures that every connection adheres to mandated standards, mitigating the risk of an employee forgetting to activate a personal VPN or using outdated credentials.

This centralization is critical in risk management.

Whether split tunneling is acceptable depends on what you can still see and enforce, not on how much bandwidth it saves.

Before enabling split tunneling on an IPsec client, decide what the non-tunneled traffic will be inspected by. NIST's telework guidance is precise about the trade: split tunneling protects only internal-resource traffic through the VPN while excluding other communications from the tunnel, which reduces load and improves efficiency but limits the organization's ability to inspect and protect traffic. If your only reason to split is that the gateway is saturated, buy gateway capacity; you are trading visibility for hardware you didn't budget.

The order of operations that holds up in practice looks like this. First, inventory what actually needs to reach internal resources — file shares, ERP, internal DNS — and confirm those destinations are covered by the tunnel's routing table. Second, decide where the non-VPN path terminates. On a corporate-owned laptop the answer should be a monitored egress, not the coffee shop's Wi-Fi. Third, configure the client so the split is enforced by policy rather than left to the user, because NIST notes the default setting that allows split tunneling is a balance between security and functionality, and a balance is a choice someone makes, not a safe harbor. Fourth, test from a hostile network: connect to a public access point, confirm internal traffic is encapsulated and confirm what the non-tunneled traffic looks like from the outside.

What goes wrong is usually lateral movement. If split tunneling lets a device reach the public internet directly while holding an authenticated tunnel to the IPsec gateway, that device is now a bridge between an uninspected network and the inside. The failure mode is quiet: no error, no dropped connection, just a machine that is simultaneously trusted and exposed. NIST says split tunneling is strongly discouraged for IPsec VPN clients because of its security complications and risks, and the reason is exactly this — the endpoint becomes the place where two trust domains meet without a checkpoint.

Where split tunneling is defensible: low-sensitivity internal resources, devices you fully manage, and a non-VPN path you can still log. Where it is the wrong choice: contractor laptops, anything touching regulated data, and any environment where you cannot answer "what inspected the other traffic?" without guessing. The cost of saying no is bandwidth and latency through the gateway; the cost of saying yes carelessly is that your inspection story has a hole shaped like every employee's home network.

Budget the breach before you argue about the tunnel; the insurance math is the part most teams get wrong.

The 2019 New Orleans ransomware incident is the cleanest case study for why VPN and access-control decisions belong in a risk conversation, not just an IT one. The city's cyber insurance policy limit at the time was $3,000,000, and the total estimated cost of the December 2019 attack was $7,200,000 — meaning the city expected cyber insurance to cover $3,000,000 of a bill roughly two and a half times that size. In the same month, Pensacola reportedly faced a $1,000,000 ransom demand. A gap of that shape is not an insurance problem; it is a configuration and recovery problem that the policy simply stops paying for.

Pensacola's later procurement is the useful template. In a 2024 RFP for risk broker services, the city specified a minimum cyber liability insurance limit of $3,000,000 and, alongside it, a professional liability per-occurrence minimum limit of $1,000,000. Read those two numbers together and the lesson for a VPN owner is that coverage limits are a floor you negotiate, not a ceiling that protects you. If your remote access design assumes the insurer absorbs the difference, you have priced the tunnel wrong.

Practically, the sequence is: map which controls your policy or a counterparty's contract actually requires — MFA on the IPsec gateway, logging retention, endpoint protection baseline — then verify you meet them, then buy the limit. Do not buy first and retrofit. A control you cannot evidence after an incident is a control the adjuster will treat as absent, and the New Orleans figures show how fast a seven-figure estimate becomes a budget crisis.

Endpoint hygiene belongs in the same conversation. AV-TEST's March–April 2024 Windows 11 home-user comparison tested 17 products, and 11 achieved the "TOP PRODUCT" rating, which requires a minimum overall score of 17.5 out of 18 points against a maximum of 18. That is a demanding bar, and it still doesn't cover the case where a user's split-tunneled path bypasses the tunnel entirely. Antivirus scores measure the endpoint; they say nothing about whether the endpoint's traffic was ever inspected. Treat the two as separate line items in the risk register, because they fail independently.