- Reuse all your existing subgraph entities.
- Increase querying speeds drastically compared to graphql-engines.
- Flexible aggregations that weren’t possible with just GraphQL.
- Analytics on protocols through Clickhouse, and more.
- Plug into BI tools, train AI, and export data for your users
Automatic Deduplication
Subgraphs natively support time travel queries. This means every historical version of every entity is stored. To do this, each row has anid, vid, and block_range.
When you update an entity in a subgraph mapping handler, a new row in the database is created with the same ID, but new VID and block_range, and the old row’s block_range is updated to have an end.
By default, pipelines deduplicate on id, to show only the latest row per id. In other words, historical entity state is not kept in the sink database. This saves a lot of database space and makes for easier querying, as additional deduplication logic is not needed for simple queries. In a postgres database for example, the pipeline will update existing rows with the values from the newest block.
This deduplication happens through setting the primary key in the data going through the pipeline. By default, the primary key is id.
If historical data is desired, you can set the primary key to vid through a transform.
Example pipeline with historical data retention
Using the wizard
Subgraphs from your project
Use any of your own subgraphs as a pipeline source. Usegoldsky pipeline create <pipeline-name> and select Project Subgraph, and push subgraph data into any of our supported sinks.