Designed to be open-source
- Self-hostable on commodity hardware
- Program-scoped ingestion via include/exclude lists
- Soon open-sourcing under AGPL-3.0 as part of the RPC 2.0
Steamboat custom indexes for expressive Solana apps
Tailored account indexes on shared infrastructure replace full-chain scans, so reads land faster with far fewer timeouts and rate-limit hits.
Indexes that learn your traffic
Steamboat watches incoming gPA requests and creates tailored PostgreSQL indexes automatically. Once an index exists, all matching queries skip the raw scans from that point on, with zero changes on your end.
- Auto-built indexes from your query patterns
- Detection on memcmp and dataSize filters
- Hot programs get dedicated in-memory indexes that stay fast under load
Indexes that learn your traffic
Steamboat watches incoming gPA requests and creates tailored PostgreSQL indexes automatically. Once an index exists, all matching queries skip the raw scans from that point on, with zero changes on your end.
- Auto-built indexes from your query patterns
- Detection on memcmp and dataSize filters
- Hot programs get dedicated in-memory indexes that stay fast under load
Query app-specific accounts data at scale
Unlike Agave, which limits apps to the validator's native indexes, Steamboat lets you query account data exactly the way your app's logic needs to display it, adapting to your frontend on the go and giving Solana state the flexibility of a database.
- Filter on any memcmp offset and length
- Compose queries across owner, delegate, and data slices
- Build views the validator's native indexes don't support
Query app-specific accounts data at scale
Unlike Agave, which limits apps to the validator's native indexes, Steamboat lets you query account data exactly the way your app's logic needs to display it, adapting to your frontend on the go and giving Solana state the flexibility of a database.
- Filter on any memcmp offset and length
- Compose queries across owner, delegate, and data slices
- Build views the validator's native indexes don't support
Heaviest calls on Solana, returned in milliseconds
Every gPA query that matches an index is served by a PostgreSQL lookup, skipping a full-chain scan. gTABO and gTABD avoid scans entirely via denormalised columns. The result is an order-of-magnitude latency drop on the heaviest reads on Solana.
- gPA average response: 12x faster than full-scan baseline
- gPA P90 tail latency: 24x lower
- gPA jsonParsed: 20x faster, no parse penalty
Heaviest calls on Solana, returned in milliseconds
Every gPA query that matches an index is served by a PostgreSQL lookup, skipping a full-chain scan. gTABO and gTABD avoid scans entirely via denormalised columns. The result is an order-of-magnitude latency drop on the heaviest reads on Solana.
- gPA average response: 12x faster than full-scan baseline
- gPA P90 tail latency: 24x lower
- gPA jsonParsed: 20x faster, no parse penalty
Built to be flexible
- Modular ingestion from Dragon's Mouth into PostgreSQL
- Four independently deployable processes
- Horizontally scalable API behind a load balancer
No premium for indexed reads
- Custom indexing without the dedicated-node cost
- Same flat rate across standard RPC, gPA, gTABO, and gTABD
- Flexible rate limits on all subscriptions
Modular infrastructure built for scale
- Spikes absorbed across independent components
- Globally distributed bare-metal clusters
- GeoDNS routing with automatic failover
Intelligent abuse prevention
- Access controls, origin allowlists and request fingerprinting
- Flexible RPS and connection limits on all subscriptions
- No tier-gated throttling that blocks legitimate workloads
Direct support channel on every endpoint
- Access to the engineering team that built the infrastructure
- Configuration help, debugging, and urgent fixes
- Reachable 24/7 across regions