Base Indexer
The Base Indexer is responsible for fetching data from the Data Source and indexing it on-chain. It acts on the endpoints configured in the previous step, pulling in fresh data at specified intervals or triggers. Chainsight supports multiple ingestion strategies for the Base Indexer, allowing a balance between cost, speed, and trust:
Cost-Friendly Backend Ingestion: Chainsight’s internal nodes fetch the data from the source API and directly push it into the on-chain indexer. This approach is simple and has minimal overhead, making it cheaper and faster for frequent updates. Pros: very low cost and quick to set up – ideal if you trust the data source or aggregator. Cons: lacks advanced cryptographic proof of authenticity, so it’s best used when cost is the primary concern and the data source is trusted.
zkTLS-Based Ingestion: In this method, an external cryptographic service/provider fetches the data and produces a zero-knowledge proof (zkTLS) that the data came from the authentic source over TLS. The Base Indexer then ingests the data along with this proof. Pros: offers high data integrity – the zero-knowledge proof ensures the data hasn’t been tampered with even though the fetch happened off-chain. Cons: slightly higher overhead and a potentially smaller range of supported sources (depending on the proof provider). This method is perfect for real-time data (like pricing) that require strong authenticity guarantees.
Distributed HTTPS Outcalls (Consensus): Here, a group of distributed Chainsight nodes each call the same HTTPS endpoint and then reach consensus on the result. If a majority (or a threshold quorum) of nodes agree on the data, that value is accepted and stored. Pros: no reliance on a single node or aggregator – this provides trust through decentralization (no single point of failure) and is great for public APIs or widely used data. Cons: it incurs a bit more latency (nodes must communicate and agree), so updates are a few seconds slower than a direct push. This trade-off is worth it for scenarios where you want to eliminate trust in any one source.
After retrieval, the Base Indexer stores time-series data on-chain. It doesn’t just keep the latest value; it retains historical snapshots in an optimized way. The retention policy is:
Minute-by-Minute for the First Hour: Every minute for the first 60 minutes, the indexer records a data point. (Total of 60 points in the first hour).
Hourly Aggregation: After 1 hour, the system compresses those 60 minute-level points into 1 hourly snapshot. Each hour yields one data point (24 points per day).
Daily Aggregation: After 24 hours, the indexer consolidates the 24 hourly points into a single daily point. Over a week you’d have 7 daily points, and so on.
90-Day Retention: The daily data is kept for up to 90 days. Data older than 90 days is discarded (unless extended storage is configured).
This tiered time-series storage means recent data is very granular (minute-level for the last hour), and older data is stored in summarized form (hourly, then daily), balancing detail and storage efficiency.
Implications: For DeFi projects and analytics, this on-chain history is extremely powerful. You can pull historical data easily – from minute-by-minute recent data to daily aggregates going back ~90 days – all from the oracle itself. This “historical on-chain ledger” of data enables advanced analyses (like calculating volatility or trends) directly in smart contracts or on-chain queries, without needing an off-chain database. Such built-in historical depth is uncommon among decentralized oracles and opens the door to richer DeFi logic and on-chain analytics.
Last updated