This is the first installment of the “Explained by…” series — engineer-written explanations of the SQD stack. DK, a core SQD developer, breaks down how the network upgrade serves data through Portal, workers, and hotblocks.
What’s SQD Network?
SQD Network is a huge and growing compute cluster consisting of 2500+ decentralized workers. Each worker contains 1 TB of storage and includes a query engine for serving network requests. The decentralized architecture means independent operators globally manage workers rather than a single controlling entity.
What’s Portal?
Portal (formerly Gateway) is the service window to the Network for users. It communicates with decentralized workers via libp2p, providing a simple and efficient HTTP API for streaming data. Portal serves as the bridge between centralized and decentralized systems.
What are Hotblocks?
Static archival data streams from the Network with high throughput, while real-time “last-mile” data requires low latency. Hotblocks store this real-time data alongside Portal and receive live updates from RPC nodes. Portal combines both sources into a unified stream — high throughput and low latency simultaneously.
Can I run my own Portal?
Yes. Users must lock SQD tokens to access the Network; more locked tokens yield higher rate limits. That said, most users find it simpler to connect through one of the existing Portal operators, which offer flexible payment plans. There is also a free public Portal available for testing before production deployment.
What data is available?
SQD Network supports raw blockchain data for 200+ networks. Users can request specific fields, apply filters, and retrieve entire blockchain histories in minutes or hours — versus traditional RPC approaches requiring days or months. Client-side aggregation handles token balances and pricing calculations. SQD provides a dedicated SDK to make that aggregation straightforward.
What exactly are we releasing at the end of 2025?
The upgrade from the centralized “v2 archives” version provides:
- Significantly larger compute cluster (petabytes of storage, thousands of CPUs)
- 10x faster Rust query engine
- Accelerated data streaming via fetch-ahead parallelism
- Built-in Hotblocks (eliminating need for external RPCs)
- Permissionless access
- Expanded datasets including full Solana and Eclipse history