History page / Light bridge: General infra with The Graph (indexing on-chain events)

[wsdt]

Based on the discussion with Ino, what the most affordable and efficient way to deploy our subgraphs for selected networks would be.
Also considering that in best case we can avoid having a mix of private and public infrastructure.

The Graph: Hosted Service

Is being discontinued, but non supported networks on The Graph Network will still be provided (for some time at least):

Conclusion: Not an option

That also explains why the Boba subgraphs are still being synced because Boba is not available on the decentralized network.

The Graph: Decentralized network

Payment methods:

• Fiat with Banxa (but just regular fiat on-ramp to buy GRT tokens)
• Crypto, billing contract living on Arbitrum

Pro's for using The Graph where possible:

• 60-98%* (official statement) lower monthly cost
• $0 infra setup costs
• 24/7 technical support by global community, superior uptime, etc.

Average cost per query is: ~$0.0002 in GRT tokens (but priced in USD).

Con's:

• Managing GRT tokens (but since billing contract lives on ARB in general, there is just one network/wallet to be funded).

Self-hosted Graph node

Con's:

• Higher infra costs
• Maintenance overhead

Pro's:

• No GRT tokens to be managed

---

Overview of what's possible

Basically, we don't need our own graph node at all for (meaning these could be paid for in GRT tokens):

• Ethereum Mainnet (L1)
• Arbitrum Mainnet
• Optimism Mainnet
• Goerli (L1)
• Arbitrum Goerli
• Optimism Goerli

I already tested these out:

But we need to deploy our own nodes for:

• Boba ETH Mainnet
• Boba BNB Mainnet
• BNB Mainnet
• Boba Goerli
• Boba BNB Testnet
• BNB Testnet

---

Just a few thoughts here:

At this point I'm questioning in general if the value generated by showing the history of bridged assets through teleportation/light bridge makes up for the tremendeous costs and overhead. Except we plan to use these graph nodes for several other contracts and infrastructure.

---

Otherwise, if we want to continue that way - I would suggest this approach:

We would be able to deploy our subgraphs on 6 existing nodes, which saves us substantial infrastructure costs.

As a consequence, I think it would be reasonable to take the pain of ensuring that the single Arbitrum billing contract is funded sufficiently (especially since its only one contract and not a contract on every respective network).

For the other networks, I see no way around of deploying our own nodes.