Why
Being able to route incoming storage requests to multiple storage nodes is a critical first step towards building our storage network.
Scope:
For this version of the endeavor, the goal is to establish router.web3.storage -- a router instance that talks to regular web3.storage (arguably we really won't exercise code paths that have to choose between nodes, but that's fine). Normal w3up clients won't talk to it by default yet, but we can use it ourselves by setting env vars for w3cli.
Steps to implement:
[x] #44
[ ] #70
[ ] Change blob protocol client to handle async response to service/blob/allocate
[ ] Implement routed a routed blob handlers (in w3up
[ ] Implement router infra (w3-router-infra)
[ ] Deploy and connect DNS
Only the design review is converted to a ticket so far cause the remaining tasks may need a slight revision when they are complete
Cost
This is a sizable endeavor, but also should be doable in 1 full sprint (or worked on by maybe 2-3 devs over 2 sprints)
Acceptance Criteria
configure a w3cli to talk to router.web3.storage
should be able to execute a blob upload that ends up on web3.storage and has a w3s.link you can download from
What Router.storage is an implementation of the W3 Blob Protocol that puts content to multiple storage nodes. A rough design for implementation can be found here -- https://github.com/web3-storage/RFC/pull/25
Why Being able to route incoming storage requests to multiple storage nodes is a critical first step towards building our storage network.
Scope:
For this version of the endeavor, the goal is to establish router.web3.storage -- a router instance that talks to regular web3.storage (arguably we really won't exercise code paths that have to choose between nodes, but that's fine). Normal w3up clients won't talk to it by default yet, but we can use it ourselves by setting env vars for w3cli.
Steps to implement:
service/blob/allocateOnly the design review is converted to a ticket so far cause the remaining tasks may need a slight revision when they are complete
Cost This is a sizable endeavor, but also should be doable in 1 full sprint (or worked on by maybe 2-3 devs over 2 sprints)
Acceptance Criteria
blocking
6