This topic provides some best practices for configuring your RPC settings to ensure a reliable infrastructure for your projects and services built on Sui.
- Use dedicated nodes/shared services rather than public endpoints for production apps. The public endpoints maintained by Mysten Labs are rate-limited, and support only approximately 100 requests per 30 seconds. You should not use public endpoints in production applications with high traffic volume.
- You can either run your own Full nodes, or outsource this to a professional infrastructure provider (preferred for apps that have high traffic). You can find a list of reliable RPC endpoint providers for Sui on the Sui Dev Portal.
RPC provisioning guidance
Consider the following when working with a provider:
- SLA and 24-hour support: Choose a provider that offers a SLA that meets your needs and 24-hour support.
- Onboarding call: Always do an onboarding call with the provider you select to ensure they can provide service that meets your needs. If you have a high-traffic event, such as an NFT mint coming up, notify your RPC provider with the expected traffic increase at least 48 hours in advance.
- Redundancy: It is important for high-traffic and time-sensitive apps, like NFT marketplaces and DeFi protocols, to ensure they don't rely on just one provider for RPCs. Many projects default to just using a single provider, but that's extremely risky and you should use other providers to provide redundancy.
- Traffic estimate: You should have a good idea about the amount and type of traffic you expect, and you should communicate that information in advance with your RPC provider. During high-traffic events (such as NFT mints), request increased capacity from your RPC provider in advance. Bot mitigation - As Sui matures, a lot of bots will emerge on the network. Sui dApp builders should think about bot mitigation at the infrastructure level. This depends heavily on use cases. For NFT minting, bots are undesirable. However, for certain DeFi use cases, bots are necessary. Think about the implications and prepare your infrastructure accordingly.
- Provisioning notice: Make RPC provisioning requests at least one week in advance. This gives operators and providers advance notice so they can arrange for the configure hardware/servers as necessary. If there’s a sudden, unexpected demand, please reach out to us so we can help set you up with providers that have capacity for urgent situations.