QuickNode

QuickNode sells blockchain node infrastructure, and IPFS pinning is one product in that catalogue rather than the business. The case for using it is that the account, the billing, and the dashboard already exist for whatever RPC endpoints the project is running on.

Its own REST API#

There are three groups of endpoints under one base — gateway management, pinning, and account usage — authenticated with an x-api-key header:

1
2
3
4
5
6
7
8

curl -X POST "https://api.quicknode.com/ipfs/rest/v1/pinning" \
  -H "x-api-key: $QN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "cid": "bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi",
        "name": "site-v4",
        "origins": ["/ip4/203.0.113.7/tcp/4001/p2p/12D3KooWExampleOriginPeerID"]
      }'

This is not the vendor-agnostic Pinning Service API, and the way it is not is instructive. The request body is the specification’s body — cid, name, origins, meta, with name required here rather than optional. origins carries the multiaddresses of nodes that already have the blocks and does the same work here as it does everywhere else. What differs is the envelope: a proprietary path, an x-api-key header where the specification says Authorization: Bearer, and no /pins root. So ipfs pin remote service add cannot point at it, no standard client library works, and porting to a provider that does implement the specification is mostly a matter of moving the same JSON to a different URL.

The account-usage endpoints returning bandwidth and storage metrics are a genuine convenience the standard specification does not cover, and they are also part of what has to be rewritten on the way out.

The structural risk is the shape of the business#

Pinning here is a secondary product of a company whose main business is node access. That is precisely the profile of the entries in the graveyard — Infura was a node provider with an IPFS service on the side, and the IPFS service is the one that was shut off.

This is not a prediction about QuickNode. It is a reason to treat the bespoke integration as a cost paid twice — once to write it, once to replace it — and to hold the second pin that the rest of this section argues for. The content identifiers (CIDs) survive either way.

Backward Lighthouse 4EVERLAND Forward