Your IPs, your regions, your bandwidth model. Already running hundreds of thousands of VPN exit IPs for operators serving real subscribers.
VPN operators spend years building IP reputation. Don't restart from zero on every cloud.
Your users in Riyadh, Istanbul, or Jakarta need a PoP they can reach in single-digit milliseconds.
Hanoi · Ho Chi Minh City · Hong Kong · Jakarta · Kuala Lumpur · Manila · Singapore
Bogotá · Buenos Aires · Dallas · Lima · Los Angeles · Miami · Santiago · São Paulo · Washington DC
30 regions worldwide, all with compute, load balancing, and IP allocation.
When your subscribers connect in one region and their traffic exits another, the path stays on Zenlayer's network, not the public internet.
VPN protocols don't fit web-shaped load balancers. Ours does what you need.
Per-GB hyperscaler egress eats consumer VPN unit economics. Pick the model that matches your ARPU.
Committed bandwidth, predictable monthly bill.
Pool capacity across multiple resources in a region.
Per-second billing, no commits.
Apply bandwidth ceilings to groups of instances (e.g. trial vs. paid tiers).
Start from a sized package or customize your own. Compute and data transfer are billed separately — a traffic-heavy region never forces you to over-provision CPU.
Every instance runs on latest-generation AMD CPUs (x86-64) with all-NVMe SSD storage.
Egress on a Tier-1 backbone trusted by enterprises and carriers worldwide. Pay metered per GB for the data you move, or commit bandwidth by the Mbps.
Compute prices shown for US/EU regions (na-east-1, na-central-2, na-south-1, europe-central-1).
WireGuard, OpenVPN, IPsec/IKEv2 — bring your image and your user plane.
REST API · Terraform · Go + Node SDKs · CLI — per-second pay-as-you-go billing.
100,000+ rounds down from the current internal count of IPv4 addresses. IPv6 excluded — address space is effectively unbounded.
Straight answers for operators evaluating where to run their VPN fleet — protocols, IPs, regions, bandwidth, and how this differs from a generic cloud.
All of them. WireGuard, OpenVPN, and IPsec run on standard Linux instances — you bring the stack, we don't wrap it in a control plane of our own. Our load balancers sit in front of your fleet and handle both WireGuard's UDP and OpenVPN's TCP.
Yes — BYOIP is first-class. Announce your own prefixes through our network and keep the IP reputation you've already built, with no renumbering. If you don't have your own space yet, you can launch on ours and migrate later.
30 regions, concentrated where subscribers actually are — the Middle East, Eastern Europe, Southeast Asia, and Latin America — all joined by a single private backbone, so traffic between your PoPs stays off the public internet.
Priced like a VPN, not a website. Choose flat-rate, a pooled Bandwidth Cluster across your whole fleet, or pay-as-you-go, and shape it with QoS policy groups. Bandwidth scales independently of compute, so a traffic-heavy region doesn't force you to over-provision CPU.
A generic VPS hands you a box and an IP off a shared, often-flagged pool. We give you carrier-grade networking built for VPN traffic: BYOIP so the reputation is yours, UDP-aware load balancing, pooled bandwidth, and a private backbone between regions. You run the VPN; we run the network under it.
Yes — our load balancer speaks UDP natively, not just HTTP and TCP. WireGuard and other UDP-based protocols balance across your instances without a TCP shim in front.
Yes — instances run dual-stack with public IPv6, so you can serve subscribers on IPv6-only mobile networks without a translation layer. IPv6 address space is effectively unbounded, so it isn't metered the way IPv4 is.