The Falcon Network is a network of fast relay backbone nodes, designed for fast dissemination of Bitcoin blocks. Fast block dissemination helps decentralization: it ensures that miners of all sizes efficiently discover new blocks, reduces orphan rates, and minimizes wasted miner effort.
The Falcon Network uses a novel technique for fast block propagation that is much faster than existing techniques.
Falcon uses cut-through routing coupled with an optimized topology. This approach is asymptotically faster than techniques based on simple block compression. These slides provide more details.
Connecting to Falcon is as easy as pointing your bitcoin daemon to one of our nodes with the --addnode and --whitelist parameters. You do not need to run any special code locally.
We would like to build up a user base with which we can communicate, however, and to that end, we are enabling the high-speed mode of Falcon on an invite-only basis.
If you want to get the full benefits of the Falcon network, please sign up! All fields in this form are optional. If you do provide an IP address, Falcon nodes will reliably maintain a connection to your servers through network failures and reboots.
You can also just connect to our nodes without going through the sign-up process, but in that case, Falcon will not employ its high-speed mode when communicating with your node.
The Fast Relay Network was a volunteer service that was operated by Matt Corallo for the benefit of the miner community for many years, but has recently stopped being maintained. Falcon and FRN have quite a few similarities, but differ on main architectural grounds.
Falcon is faster than the FRN. FRN and Falcon differ fundamentally on one critical issue: how they achieve their speed gains. Falcon uses cut-through routing, which can ensure that data is propagated as soon as any portion of a block has been received. In contrast, FRN is based on a store-compress-and-forward architecture.
Falcon uses an optimized topology. FRN uses a static, manually-crafted topology, while Falcon's topology is modified dynamically to minimize latencies.
Falcon requires no set-up. FRN requires locally installing and running special decompression code, while connecting to Falcon just requires passing an --addnode option to your bitcoin daemon.
Falcon is fully supported, while FRN is no longer maintained.
Overall, there is great value in having multiple relay backbones, operated by different entities. We encourage miners to connect to both FRN and Falcon to ensure the best possible service.
Only subscribed nodes get the benefits of our high-speed transport, so you will not receive the same high-speed benefits that other nodes do unless you subscribe.
Yes, we do not reveal the IP addresses of the nodes connected to our nodes.
By making the network faster, Falcon reduces the adverse effects of larger blocks. It reduces orphan rate, reduces wasted miner effort, and improves fairness and decentralization. As a result, Falcon enables blocks to be made bigger than they could maximally be under the regular Bitcoin peer-to-peer network or the old "fast relay network."
Some of the arguments made in the block size debate are not driven by the performance of the consensus layer (e.g. being able to support severely resource-constrained nodes); Falcon does not impact that aspect of the debate.
You should connect to the Falcon node that is closest to you in terms of ping distance. There is no advantage to peering with as many relay nodes as you can; in fact, that might result in higher bandwidth usage.
Falcon does not fully verify the data it is relaying. You should always check every block with your own validator before building on it, as you would for any other block from the Bitcoin P2P network.
The relay nodes send blocks immediately to peers that they believe do not have that block, so you may see blocks being sent which you already have. If this is a problem for you due to bandwith issues, you should reconsider your bandwith provisioning.
Falcon is not a replacement for having peers on the standard P2P network.
Falcon is blazingly fast. The network takes advantage of network-wide parallelism, and ensures that blocks are transmitted as they are being actively received by the network. This means that block transmissions can be completed within tens of milliseconds after the last bytes of the block are received by the network.
Falcon is experimental software. We frequently deploy new software updates, some of which might require a disconnect/reconnect. If you are subscribed, Falcon nodes will automatically reconnect back to your node.
The Falcon service is entirely self-supported by the research initiative that developed the technique. Please consider becoming a sponsor.