Flaem

IP transit, IPv4 leasing, ASN and BGP sold like any other service

Once you run your own network, companies buy more than servers from you: internet access, IP addresses and help getting a network number of their own. Flaem runs these services from the order in your store, through the BGP session and the RIPE database object, to the invoice and the renewal.

Upstream provider
your BGP session to the world
Your network
prefixes, links and nodes
Flaem: store, invoices, RIPE
BIRD 2 on the node
01Services

Four network services in your store, from order to running

Each one has its own offer, price and billing period, just like a game server or a VPS. Once paid, Flaem sets up whatever the network side needs, and when a payment is missed it suspends the service instead of deleting it.

From order to running
01

IP transit

A company with its own network buys internet access over your links and announces its addresses through you.

  1. Order

    The client picks transit from your offer and pays for it like any other service.

  2. BGP session

    The session with the client comes up on BIRD 2 on the node, with a route filter that belongs to that session only.

  3. RPKI and announcement

    The client's prefixes go through your RPKI validator. A route marked invalid goes no further.

  4. when the client does not pay

    Suspension

    When a payment is missed, Flaem stops announcing the client's prefixes and resumes once they pay.

02

GRE tunnel

The client receives traffic for their addresses over a tunnel, including on a machine that sits outside your network.

  1. Order

    A tunnel is a product of its own in the store, with its own price and billing period.

  2. Built on the node

    Once paid, the node builds the GRE tunnel itself, with nobody logging into a router.

  3. History

    Traffic in bits and packets, connection age and tunnel restarts stay in the statistics.

  4. when the client does not pay

    Suspension

    An unpaid tunnel is suspended the same way as any other unpaid service.

03

IPv4 and IPv6 leasing

A company that needs addresses leases them from you instead of obtaining a pool of its own.

  1. Pool on offer

    You set aside a pool, a prefix or a session for leasing and sell it with renewals.

  2. RIPE object

    The subnet assignment and the route object go into the RIPE database straight from the panel.

  3. Assignment

    The block is tied to the account of the client or sub-operator leasing it.

  4. when the client does not pay

    Suspension

    The announcement is suspended and the block only returns to the pool after the number of days you set.

04

ASN registration

A company that wants a network number of its own handles the RIPE NCC formalities through you as its sponsoring LIR.

  1. Request

    The client fills in the ASN request in their panel.

  2. Documents

    They attach the company documents in the same place, with no email back and forth.

  3. Status

    They follow the request the same way they follow an order.

  4. BGP session

    With the number assigned, they order transit and get a BGP session.

02Prefix path

The path of a client prefix, from the panel to your upstream

A prefix is a block of addresses that a network announces to the rest of the internet. Before a client's addresses work, they pass through five places, and you set up every one of them in the same panel.

  1. 01

    Client account

    The prefix belongs to the client's account and to their transit or leasing service, so it is always clear whose it is and why it exists.

  2. 02

    RIPE database

    The panel writes the route object, the public record of which network number announces this prefix, and checks that it carries the right ASN and your maintainer.

  3. 03

    RPKI validator

    Your validator, Routinator for example, checks the prefix signature. Flaem blocks an announcement marked invalid before it reaches the router.

  4. 04

    BIRD 2 on the node

    The router on the node accepts the route only if the client session's filter lets it through. Flaem generates and loads the BIRD configuration fragment itself.

  5. 05

    Upstream provider

    The prefix goes out to the world over your session with the upstream. Session state, drops and prefix counts stay in the statistics.

03BGP sessions

BGP sessions with clients, sub-operators and your upstream

BGP is how networks tell each other which addresses they serve. Flaem runs those conversations on BIRD 2 in three directions, and every session has an owner, a route filter and a history.

Upstream provider
blackhole and FlowSpec
Your network
BIRD 2 on the node
Clients with their own ASN
Sub-operators
route filter and RPKI
route filter and RPKI
route filter and RPKI

Clients with their own ASN

A company that bought transit has its own session with you. Its prefixes pass that session's filter and the RPKI validator, so nobody else's addresses go out to the world in your name.

Upstream provider

This session carries your prefixes and your clients' prefixes out to the world. Blackhole and FlowSpec rules go the same way when an attack has to be stopped before it reaches your link.

Sub-operators

A smaller operator reselling your links and addresses has a separate session, and the space it leases is recorded in the RIPE database as an assignment to it.

Networking module: AS number, prefixes, BGP sessions with RPKI filters and provider-side protection

BGP session history

A state timeline, drops, prefixes received, accepted and sent, and the last error. You see when a session came up and why it went down.

Tunnel history

Traffic in bits and packets, connection age and restarts of every GRE tunnel, next to the session carrying its traffic.

04IP addresses

IPv4 and IPv6 addresses: pools, fixed client addresses and leasing

Every address in your network has a state in the panel: free in a pool, assigned to a service, tied to an account or leased to another company. That is why the same address never ends up on two machines.

Diagram of an address block. Colour shows the state of each address.

And on every node

Addresses without cloud-init

The node hands out addresses over DHCP and IPv6 router advertisements on the bridge, pinned to the MAC address, so Windows, BSD and custom images get a network without configuration. Before handing out an address it checks that nothing answers on it.

VPS without its own IPv4

You can also sell a VPS without its own IPv4 address, as a cart option of the same plan.

Reverse DNS

The client sets rDNS for their addresses in the panel, and your PowerDNS serves the reverse zones.

Cloudflare as DNS

If you do not run your own DNS servers, your clients' domain zones live in Cloudflare.

05ASN

ASN registration for clients: request, documents and status in the panel

An ASN is the number a company's network goes by on the internet. RIPE NCC assigns one to a network that is multi-homed and has its own routing policy, and the company signs an agreement with a sponsoring LIR or with RIPE NCC itself. If you are an LIR, Flaem turns the email exchange into an order the client follows in their panel.

Client
You as the sponsoring LIR
RIPE NCC
  1. Request

    The client fills in the ASN request in the panel.

  2. Documents

    They attach the company documents to the same request.

  3. LIR request

    You file the request with RIPE NCC on the client's behalf.

  4. Review

    RIPE NCC checks that the network is multi-homed and has its own routing policy.

  5. Number

    RIPE NCC assigns the ASN.

  6. Status and BGP session

    The client sees the assigned number in the panel, orders transit with it and gets a BGP session.

06DDoS protection

DDoS protection at your provider, configured by you

A DDoS attack is a flood of traffic meant to cut a server off the internet. Flaem connects the protection you already have at your provider, sets it up for every service and shows clients the attacks on their addresses.

Internet
attack trafficplayers and clients
Scrubbing at the provider
Path.net, OVH, Cloudflare, NeoProtect
dropped
clean traffic
Your node
kernel filter
Client service

Providers you connect in the panel

Provider keys go into the admin panel, and you switch each integration on separately.

Path.netA game-profile filter on the server's main port, a plain filter on additional ports and on ports the client adds, rules for single addresses and attack history in the panel.
OVHGame profiles, Minecraft Bedrock included, Edge Firewall rules and mitigation profiles on server ports. Ongoing attacks show up in the panel.
Cloudflare Magic TransitCloudflare announces your prefix over BGP, scrubs the traffic in its network and hands it back over a GRE or IPsec tunnel. You switch the Cloudflare announcement on and off from the panel.
NeoProtectDDoS protection as a provider service, connected to your addresses and run from the same panel as the other providers.
07Filters

Filters assigned to the game, not to the client

A game server client should not need to know what a filter profile is for their server to be protected. You set filters once, for port ranges, and every new server lands behind the right filter on its own.

065535
25000Minecraft: game profile26000
server main portport added by the clientother ports: plain filter

A port range for one game

You assign a game filter to a port range, for example Minecraft only on ports 25000 to 26000, and register that range with your protection provider. Minecraft servers only get ports from that range, so they always sit behind the Minecraft filter. The server's main port has the game profile and ports the client adds get the plain filter.

When the filter is not enough

closer to your node
  1. Kernel filter on the node

    on your node

    The node drops junk traffic before it reaches the servers, answers server list queries from memory and limits packets per address. You switch it on and off in the panel.

  2. Provider filter

    at the protection provider

    The game profile or the plain filter runs in the provider's network, so attack traffic never takes up your link.

  3. FlowSpec

    at the upstream

    The upstream cuts only the chosen traffic, for example UDP to a single port. Everything else to that address keeps flowing.

  4. One-click blackhole

    at the link provider

    When an attack is bigger than the link, the node announces the attacked address with the provider's blackhole signal for the time you set and withdraws it afterwards. Every use goes into the audit log.

closer to the attack source

Blocklists

Traffic from addresses and networks on Spamhaus and DNSBL lists is filtered before it reaches services. You check the same lists before handing an address to a client.

Attack reports in the client panel

Clients see attacks on their addresses in their own panel, so they do not need a ticket to find out why the server slowed down for a moment.

08Network insight

Looking glass, link quality and traffic analysis in one panel

Some network tools you show to clients and some stay with you. Both live in the same panel, with no separate monitoring system.

Your client sees

Looking glass and latency test

A public Flaem page, like the status page, or an element to embed on your website. The client measures the route and latency to your network before buying anything.

You see

Continuous link quality

Latency, packet loss and jitter every few seconds to provider gateways, other nodes and public points, a traceroute every few minutes and route change history for every provider.

Your client sees

Attacks on their addresses

Reports of attacks on the client's addresses sit in their panel, with no question to support.

You see

sFlow and NetFlow

Traffic analysis from your switches and routers, broken down by application, port and protocol.

Your client sees

Reverse DNS and addresses

The client sets rDNS and sees which addresses belong to their account and their services.

You see

On-demand traffic capture

You pick a machine or an interface, a filter and up to 60 seconds of capture. You get an encrypted file for a short time and a summary: sources, protocols and packets per second.

Your client sees

Capturing their own VPS traffic

When you allow it, the client captures their machine's traffic themselves instead of asking for it in a ticket.

You see

Exposure scanner

A periodic scan of your pools, within your range only: open attack reflectors (DNS, NTP, memcached, SSDP) and exposed services (RDP, SMB, Redis, Elasticsearch). You decide whether to message the client.

client machinemany targets at once

Outgoing attacks from client machines

Sometimes a client's machine attacks others because someone took it over. The node counts packets per second, UDP share and the number of targets for every machine, so an outgoing flood and SSH or RDP password guessing reach your abuse cases before the link provider cuts your address.

  • packets per second
  • UDP share
  • number of targets
You choose the response
  1. watch
  2. alert
  3. packet limit
  4. suspension
09Switches

Switches and node networking from the panel

If you run your own network gear, you do not log into every switch separately. Ports, VLANs and node networking change in the same place where you sell services.

port with a client machineport disabled for a missed paymentclient VLANfree port

Switch ports

You see your switch ports, the machine connected to each port, port traffic and VLANs. On abuse or a missed payment you disable a port with one click instead of logging into the switch.

Node network changes with automatic rollback

You change a bridge, a VLAN, NIC bonding, MTU or an address in the panel, on many nodes at once if you like. A mistake will not cut a node off for good: if it does not confirm connectivity with the control plane after the change, it goes back to the previous configuration on its own.

  1. 01

    Change

    You add a bridge, a VLAN, NIC bonding, MTU or an address.

  2. 02

    Diff preview

    Before saving you see exactly what will change.

  3. 03

    Confirmation

    The node has a set time to confirm connectivity with the control plane.

time to confirm
Connectivity confirmed: the new configuration staysNo confirmation: the node goes back to the previous configuration
10Questions

Questions about networking, BGP and ASN

If your question is not here, write to us. We will show you the answer on your own network.

In the network of your protection provider: Path.net, OVH, Cloudflare Magic Transit or NeoProtect. Flaem sets up filters for services, registers port ranges with the provider and shows clients reports of attacks.

BIRD 2 on your nodes. You set up sessions with clients, sub-operators and your upstream from the panel, every one has a route filter and your RPKI validator checks the prefixes.

Yes. The ASN request on a company's behalf is filed by a sponsoring LIR, a member of RIPE NCC. Flaem handles the request, the documents and the status on your client's side.

They do not have to. You set filters for games and port ranges, and the client is protected from the first start of their server.

The announcement is suspended and the block only returns to the pool after the number of days you set. Until then it goes to nobody else.

The node has a set time to confirm connectivity with the control plane. If it does not, it returns to the previous configuration on its own.

The customer network, meaning BGP sessions, prefixes, tunnels and transit, is one option at $15 a month, and DDoS protection costs $10 however many providers you connect. The number of sessions, prefixes and clients does not change the amount.

We will show BGP, RIPE and DDoS protection on your network

Tell us what address space you have, who you run BGP sessions with and who protects you from DDoS. We will prepare a demo built around your setup.