Create isolated Layer 2 networks between your servers in the same region. Internal traffic stays on our switching fabric, never touches the public internet, and incurs no bandwidth charges. Attach VPS, dedicated, and bare metal servers to the same private network.
Private networks are implemented as isolated VLANs on our switching fabric. When you create a private network, we allocate a VLAN ID and configure it across all top-of-rack switches in the region. Servers attached to the network get a virtual interface on that VLAN.
Traffic between servers on the same VLAN stays entirely on our internal fabric. It never leaves the data center, never traverses a public router, and is never visible to other tenants. Each private network is cryptographically isolated — there's no way for traffic to leak between VLANs.
You assign private IP addresses from any RFC 1918 range (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16). We don't enforce a specific subnet — use whatever addressing scheme fits your architecture.
The most common use case is separating application tiers. Web servers face the public internet and communicate with application servers over a private network. Application servers communicate with database servers over a second private network. The database servers have no public IP addresses at all.
This reduces your attack surface — database ports are never exposed to the internet. It also eliminates bandwidth charges for internal traffic, which can be significant for data-intensive workloads.
Private networks aren't limited to a single instance type. You can attach VPS instances, dedicated servers, and bare metal machines to the same network. This lets you build architectures that combine different compute tiers — a bare metal database server communicating with VPS application servers, for example.
GPU servers can also be attached to private networks. This enables training pipelines where data preparation runs on standard VPS instances and feeds data to GPU servers over the private network without bandwidth charges.
Internal traffic runs at the full port speed of the server (10 Gbps). Jumbo frames (MTU 9000) are supported for workloads that benefit from larger packet sizes, such as NFS and iSCSI. There's no limit on the number of servers attached to a single network.
A single server can be attached to multiple private networks simultaneously. Each network appears as a separate virtual interface on the server. This enables network segmentation patterns where different applications or tiers communicate over isolated networks.
Free bandwidth. 10 Gbps. API managed.