You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
researching the necessary components for the indexer
how it should look like — architecture & stack -wise
creating a tracker with subtasks
Some notes (not rules, more like, ideas to possibly follow or stray away from)
I don't see a reason for the indexer not to be a separate component from the server — i.e. an indexer should be able to:
Start up & connect to the chain
Query all existing storage providers
(connection magic)
Ask them all what do they have on the network
This implies that the server will need to be extended with an API to share what they're storing.
Some questions:
Having the SPs have PeerIDs allows them to change IP addresses with less effort, as the discovery requires bootstrap nodes, we should also have an answer/reasoning to the question — who should be a bootstrap node?
Some considerations:
Storage providers may have an indexing service but they are not required to.
Storage providers may refuse connection to an indexing service implying they’re “semi-private”.
Storage providers may (i.e. ideally) gossip information between each other too.
Indexing services gossip all information between them.
Library comparison
libp2p
iroh
waku
(notes)
docs
🟢
🟡
🔴
its really hard to figure out what waku has or not
prod ready
🟢
🟡
🟢
iroh may be but there are no big projects using it
language support
🟢
🟢
🟡
rust & js (rust is unclear for waku, there are bindings but thats it)
protocols
🟢
🟡
❓
waku doesnt even advertise them
maturity
🟢
🔴
🟡
some people on the iroh team were part of the filecoin team, others have worked on libp2p directly & iroh's original product was an ipfs node — they know what they’re doing
protocols
🟢
🟡
❓
http & websockets
discovery
🟢
❓
❓
iroh definitely has this but not quite clear, much less for waku
The text was updated successfully, but these errors were encountered:
polka-index
— not to be confused with the crate in the repoThe
polka-index
is a network-indexer, similar to the IPNI — https://docs.filecoin.io/storage-providers/architecture/network-indexerThis task consists of:
Some notes (not rules, more like, ideas to possibly follow or stray away from)
I don't see a reason for the indexer not to be a separate component from the server — i.e. an indexer should be able to:
This implies that the server will need to be extended with an API to share what they're storing.
Some questions:
Some considerations:
Library comparison
The text was updated successfully, but these errors were encountered: