I’m pretty new to IPFS but like it a lot. Now, say I wanted to help out the network by having it use a Terabyte or so of the storage on my note to cache much requested data that’s not in the immediate vicinity, node-wise speaking, pin such data even in an automated fashion. How would I go about that?
I guess the aproach would vary depending on how you want to help. If you want to be sure precious data don’t vanish have a look to colab.ipfscluster.io . If you want to accelerate popular data, just pin it. You could listen for DHT traffic to see what data are highly hot. You can ask for providers to count them to tell which have a low number of providers per requests ratio. You could also infer where are the providers to pin in priority the data with no provider in your area. Be aware that when requesting and downloading the data, you will clutter the network, not help it. So to maximize effectiveness, pin data that will be requested in the future, not efemeral data that nobody will ask for a long time. I guess the longer a CID has been actively requested, the higher the chance it will be requested again.
Thanks for the info, so except for colab.ipfscluster.io, ther eis nothing ready-made, yet? That’s a bit of a shame. I thought something like that should really come with basic IPFS-tools.
Well, just by being alotng lived node, you keep a large routing table and help maintain the DHT, helping users locate content and peers faster. You can also configure your node to relay connections and help users behind NAT. These tasks are bandwith intensive more than space intensive, but is is super easy to do.
There is no vanilla way to automatically pin content you don’t explicitely chose, but should PL implement this, how would you decide on what criteria you chose content? Colab clusters just help to propose data worthy of maintaining.