Network and backbone

One edge for every plan: L3/L4 scrubbing, IPv6 /64, editable PTR, BGP communities on request. BTC, XMR or USDT settle the same invoice.
DDoS
Layer 3/4 filtering runs continuously at roughly 1 Tbps of edge capacity; an optional L7 profile exists for HTTP floods. Both ship inside the plan price - protection here is not a second line item. A $6 Core paid in Monero and a dual-EPYC box paid in USDT ride the same scrubbing path. Selling "DDoS protection" as an upsell is a business model we declined. Indicative round-trip times live in the looking glass.
The traffic classes the edge sees, and where each is handled:
- L3 floods - ICMP floods, IP fragmentation, amplification reflections (NTP, DNS, Memcached): dropped at the scrubbing edge, before the rack.
- L4 floods - SYN floods, TCP state exhaustion, UDP storms at the port: filtered at the same layer; the port stays reachable to legitimate handshakes.
- L7 exhaustion - HTTP request floods, slowloris, API hammering: the optional L7 profile plus your own application limits (nginx rate zones, game-server tick guards). No scrubber thinks for your application.

VPS from $4.80* — Core on the annual term (list $6/mo), one invoice for 12 months, no auto-charges. Dedicated with IPMI from $39.20/mo. Crypto invoice, no KYC.
Launch nowIncluded here versus the add-on model
The market mostly prices protection as an accessory. The honest comparison, source-linked and dated:
| Provider | Model | Entry price for protection |
|---|---|---|
| VPSbit | L3/L4 at ~1 Tbps class included in every VPS and metal plan; L7 profile optional | $0 - it ships with the plan |
| BuyVM (Frantech) | Per-IP protection add-on, 700M+ packets-per-second filtering class on 3500 Gbps+ capacity | $3.00/mo per protected IP (checked 2026-10-25) |
Both models work; they just bill differently. The difference that matters at checkout: a $4.80* Core here already sits behind the scrubber, and a ten-box fleet pays for capacity once through the plan, not per IP. For attack taxonomy background, the Cloudflare learning center keeps a current classification of DDoS types.
What filtering does and does not stop
Volumetric floods - SYN, UDP amplification, reflection - are absorbed at the edge before they reach your port. Application-layer exhaustion (HTTP request floods, slowloris) needs the L7 profile plus your own rate limits; no scrubber replaces sane nginx or game-server limits. During an attack, traffic is announced, cleaned and forwarded; legitimate packets occasionally pay a latency tax measured in milliseconds. The looking-glass probes stay live throughout, so you can watch what an incident actually does to RTT instead of reading a status rhyme.
IP and DNS
Every KVM instance carries one IPv4 plus a /64 of IPv6. Reverse DNS is yours to edit from the panel the moment provisioning finishes. Metal can request a /48. PTR records are mail and VPN hygiene, not an identity field, and we are not in the domain business.
BYOIP
Bring-your-own space mounts on the top two dedicated SKUs, Summit and Monolith. The mechanics are an LOA plus your ASN and a few days of coordination with the facility - it is a routing change, not a checkbox in the cart.
Transit and peering
Amsterdam terminates on AMS-IX and DE-CIX. Zurich on SwissIX and DE-CIX. Frankfurt on DE-CIX, Luxembourg on LU-CIX with AMS-IX reach. Reykjavik on FICIX, Helsinki also on FICIX, Stockholm on NetNod, Oslo on NIX. Asia closes the ring at SGIX in Singapore, HKIX in Hong Kong, JPIX in Tokyo and TWIX in Taipei. Per-facility uplink figures sit on each location page, because that is where they are contractually true.

Latency, honestly measured
Indicative RTT to Amsterdam from our own probes: sub-millisecond in Amsterdam itself, roughly 8 ms from Frankfurt and Luxembourg, 9 ms from Zurich, 18 ms from Reykjavik, 20-25 ms from Stockholm, Helsinki and Oslo. The long legs are physics: about 180 ms to Singapore, 210 ms to Taipei, 220 ms to Hong Kong and 230 ms to Tokyo. Numbers are indicative, not an SLA - measure from your own users with the looking glass.
The whole matrix in one table - facility, exchange, uplink class, RTT to the Amsterdam core (catalog figures, checked 2026-10-25):
| Location | Exchange | Uplink | RTT to AMS |
|---|---|---|---|
| Netherlands (AMS) | AMS-IX / DE-CIX | 40 Gbps | < 1 ms |
| Germany (FRA) | DE-CIX Frankfurt | 100 Gbps | ~8 ms |
| Luxembourg (LUX) | LU-CIX / AMS-IX | 40 Gbps | ~8 ms |
| Switzerland (ZRH) | SwissIX / DE-CIX | 100 Gbps | ~9 ms |
| Iceland (RKV) | FICIX | 10 Gbps | ~18 ms |
| Sweden (STO) | NetNod / DE-CIX North | 40 Gbps | ~20 ms |
| Finland (HEL) | FICIX | 20 Gbps | ~25 ms |
| Norway (OSL) | NIX | 20 Gbps | ~25 ms |
| Singapore (SIN) | SGIX | 100 Gbps | ~180 ms |
| Taiwan (TPE) | TWIX | 40 Gbps | ~210 ms |
| Hong Kong (HKG) | HKIX | 40 Gbps | ~220 ms |
| Tokyo (TYO) | JPIX / JPNAP | 100 Gbps | ~230 ms |
Cross-checks against public exchange data live on PeeringDB; the contractually true per-facility figures sit on each location page.
Facility grade
The sites run Tier III or better with 2N or N+1 power across twelve elite locations in Europe and Asia. Uplinks run from 10 Gbps in Reykjavik through 40 Gbps in most markets to 100 Gbps in Zurich, Frankfurt, Singapore and Tokyo. Facility tier and law per city are laid out in the location table on the locations index.
Port sizes: set by the SKU, not the city
The port you get follows the plan, and the plan prices the same in every facility. KVM VPS lines run unmetered from 1 Gbps at the floor to 10 Gbps at the top of the ladder. Bare metal starts at 1 Gbps on the entry Ryzen 9 9950X and reaches 25 Gbps on the dual-EPYC Monolith, IPMI included at every rung. None of these are burst ratings that evaporate after five minutes; the meter sits on the uplink and the same table applies in Reykjavik as in Dubai. When a workload outgrows the port, the honest fix is the next SKU, not a faster billing tier.
Choosing a city without superstition
Pick the rack for two measurable things: distance to your users and the law that governs the building. Everything else - the flag, the fiat currency, the coin at checkout - changes nothing about the packet path. An archival node pulling terabytes over a weekend wants a short hop to the machines that feed it. A Monero node or a wireguard endpoint wants jurisdictional separation from the people it inconveniences, which is what dedicated servers with out-of-band IPMI are for. Both orders land on the same invoice page and settle in the same nineteen coins, so geography and payment stay independent decisions.
Redundancy across facilities
Twelve sites in Europe and Asia also mean you can run hot spares without a second vendor. Panel snapshots and a Borg or Restic target replicate cleanly between facilities; a DNS TTL is the only dial that decides how fast a failover becomes visible to users. The locations index publishes the facility tier, uplink class and governing law per city, so a redundancy plan can be written from numbers instead of sales adjectives - and re-checked after every refresh of that table.
What the coin does not change
Paying in XMR, BTC or USDT does not reroute your traffic. ASN, scrubber profile, IPv4 and port speed are properties of the plan; the coin only settles the USD figure. Send the asset the invoice names, on the network it names.
BYOIP?
Summit and Monolith SKUs. LOA plus ASN, a few days of routing coordination.
Is DDoS extra?
No. L3/L4 is included on every VPS and every metal box; L7 is an option, not a surcharge line.
How big an attack does the included filter take?
The edge is provisioned at roughly 1 Tbps of scrubbing class, which covers the volumetric floods that reach a single tenant's port; sustained siege-scale campaigns are an upstream matter and the honest answer is a looking-glass watch, not a number on a badge.
Does L7 filtering run by default?
No - L3/L4 is always on; the L7 profile for HTTP floods is an option you switch on, and your own rate limits do the last mile either way.
What happens to my box when a neighbour is attacked?
The scrubber works per-destination, so a flood aimed at a neighbour is cleaned before it reaches either rack; your traffic pays at most a milliseconds-class latency tax during the incident, visible on the pulse charts.
Can I watch latency myself?
Yes - the looking glass publishes indicative RTT from all twelve facilities, and the pulse charts on this page track the same probes.
Which coin should the invoice name?
Whichever you hold: XMR when the chain should not follow you, USDT-TRC20 when the invoice must track dollars, BTC when the payer expects it. The network never sees the answer.
Ready to launch?
Build the box — VPS or bare metal — create the password, pay the invoice that follows.
Network
Edge RTT
Uptime · 36h
Complaints · 36h