Specialized Subset(s) of the Public IPFS Swarm

[HugoBeluga]

We are trying to build a more stable and reliable network in the ipfs public network to improve the efficiency of the IPFS network for online services, which we call "Core Node Network". The Core Node Network will divide IPFS into core and lite nodes, optimizing the access efficiency of online and offline data, as well as the optimization of data volume storage for different scenarios. At the same time, the Core Node Network is fully compatible with the IPFS public network for full connectivity. We will have more requirements for the nodes joining in the heavy node network (we call them core nodes), such as maintaining stable service delivery and operation in line with expectations. This allows the core node network to provide more stable and fast file services and more reliable file backups. In addition, we are currently designing incentives for core nodes, permission control system based on core node network and blockchain, and data privacy security design, etc. Some of them have been implemented, but of course there is still a lot of work in progress. Now we are willing to contribute the design ideas of this network and discuss them with everyone. Any criticism, suggestion or approval will be of great help to us, and if we mention anything inappropriate or have any problems, your suggestions and guidance will be most welcome and appreciated. We are also looking forward to receiving the community's approval and submitting more detailed designs for discussion with you to promote the development of the IPFS community and technology applications.

[guillaumemichel]

It seems to me that this would be a perfect application for a Composable DHT. I will link the IPFS Thing presentation of the Composable DHT concept once it is out.

The Core Node Network can be a sub DHT within the main DHT. Clients could configure Core Nodes within their routing table in order to meet the goals that you have, while still being connected to the main (Banana?) DHT. It will then be easy to measure any benefit in terms of content routing latency/network resiliency etc.

[HugoBeluga]

@guillaumemichel Thank you very much for your suggestion, Composable DHT is indeed very suitable for The Core Node Network, we will try to use Composable DHT in the design and implementation later. Is there any more information about Composable DHT that can be provided to me? I can't find it on the internet

[guillaumemichel]

@HugoBeluga the Composable DHT is a concept that was discussed at IPFS Thing 2023 [talk]. We don't have any timelines for it yet.

[HugoBeluga]

@guillaumemichel I would be happy to participate in the discussion and implementation of the Composable DHT if needed

[guillaumemichel]

Great, thank you! We still need to figure out some stuff internally, but I'll get back to you once we get to it!

[MichaelWongg]

looking forward to further progress！