Guides
Running the infrastructure, not just calling it
11 articles on getting, choosing, running, scaling and watching a node — general where the problem is general, and per chain where it is not. Written for the decisions that come before picking a provider from the rankings, and the ones that come after deciding not to.
Ethereum
BSC
Solana7 articles for Solana · page 1 of 2
- Choosing· All three chains
Getting a dedicated node
“The free endpoint is not enough any more. What now?”
What a public endpoint actually gives you, what changes when you pay for a dedicated one, and how to tell which of the two your application needs — using the limits providers publish and the behaviour measured here.
- Clients· All three chains
Why the client mix matters
“Does it matter which client everybody else is running?”
What a supermajority client bug costs at 33% and at 66%, why the published diversity figures disagree with each other by more than they agree, and what any of it changes for an application rather than a validator.
- ClientsSolana
Solana clients
“Agave, Firedancer, Frankendancer, Jito — which is which?”
Three implementations and a fourth thing people call a client, why an RPC node is a different job from a validator, and why getVersion cannot tell you which one you are talking to.
- Running a nodeSolana
Running a Solana RPC node
“Is a Solana RPC node the same as a validator?”
No — and sizing one from validator requirements is the mistake that produces a node permanently behind the cluster. Hardware quoted from Anza, why the accounts and ledger need separate devices, and what account indexes cost.
- Running a node· All three chains
Exposing RPC without giving it away
“Can I just open the port?”
No. An open JSON-RPC port is an unauthenticated public interface to your node with no rate limit and no accounting — what to bind, what to put in front of it, and which namespaces never belong on a listener anybody else can reach.
- Scale· All three chains
Serving RPC at scale
“One node works. How do I serve real traffic?”
Load balancing that understands RPC, routing methods to the node that can answer them, what is safe to cache and what is not, and why failover between two providers is a different exercise from failover between two of your own nodes.