Network · Ashburn

Where the packets actually go

Each node is a pair of hosts with local NVMe storage and a 1 Gbit/s uplink. Nodes do not share disks, and a world never lives on more than one of them at a time.

Latency from Ashburn

Round-trip estimates between this node and every other node in the network. Round numbers, not marketing ranges: they are recalculated from coordinates when the site is built.

NodeCountryLatencyState
US Ashburn this node United States 1 ms Operational
NL Amsterdam Netherlands 122 ms Operational
FR Paris France 122 ms Operational
DE Frankfurt Germany 128 ms Operational
SWE Stockholm Sweden 130 ms Operational
CHZ Zurich Switzerland 131 ms Operational
FI Helsinki Finland 136 ms Operational
EE Tallinn Estonia 136 ms Operational
LV Riga Latvia 139 ms Operational
PL Warsaw Poland 141 ms Operational
LIT Vilnius Lithuania 142 ms Operational
RU Moscow Russia 153 ms Operational
RU-MSK Moscow Russia 153 ms Operational
RU-02 Moscow Russia 153 ms Operational
RU-SH Moscow Russia 153 ms Operational

What a node is

Compute

Two hosts, each with a recent-generation 8-core CPU and ECC memory. Game instances are pinned to dedicated cores where the host layout allows it, so a noisy neighbour is a scheduling bug rather than the business model.

Storage

Two NVMe drives per host in a mirror, plus a separate backup volume. World files live on NVMe; backups are written to the backup volume and replicated to a second machine in the same region.

Uplink

1 Gbit/s committed, 10 Gbit/s burst, with two transit providers and a local internet exchange. Game and voice traffic is shaped separately so a backup restore cannot starve a server.

Specifications

ItemValue
Node locationAshburn, United States
Time zoneAmerica/New_York
Support hours08:00–22:00 ET
Uplink1 Gbit/s committed, 10 Gbit/s burst
StorageNVMe mirror, separate backup volume
IPv4 / IPv6Dedicated IPv4, native IPv6 on request
Backup window02:00–05:00 local time

DDoS filtering

Filtering happens in two places. The transit provider drops volumetric floods before they reach the node, and the node applies per-protocol rules for Minecraft, Source, Steam and the voice relay. A filtered attack does not take the node offline, but it can raise latency for a few minutes while the rules settle.

What we do not promise

No provider can absorb an unlimited flood. If an attack exceeds the uplink capacity of the region, traffic is blackholed for the duration rather than letting it degrade the other customers on the node. We tell affected customers when that happens.

Moving between nodes

A world can be moved to another node in the same region without a new address. Cross-region moves are a manual job because the data has to travel: we copy the world, verify the checksum, and switch the DNS record when the copy is confirmed. Usually a single evening for worlds under 50 GB.

Tell us what you are running

Write a few lines about the game, the player count and the region you need. We answer within one business day with a concrete node and a date.

Request access