Introduction
A comprehensive introduction to Sorada, BARA SVM's archival RPC data indexing and read-optimization solution
Last updated
A comprehensive introduction to Sorada, BARA SVM's archival RPC data indexing and read-optimization solution
Last updated
Blockchain networks aggregate data through blocks, transactions, and metadata.
Querying APIs are openly available and well-documented for fetching these data directly from the network.
This may work for small decentralized applications (DApps) with minimal historical data needs, but for bigger historical data requirements the baseline APIs fail to meet 2 main business requirements:
Performance
Scalability
Companies have barely any control over the response times and the data operation optimizations of the baseline querying API endpoints provided by the network.
This is critical for business requirements because performance greatly impacts .
Similar to performance, companies cannot optimally adapt to the demands of their users with elastic provisioning due to the limitation of the baseline Querying API endpoints provided by the network.
This is another critical business requirement because scalability goes beyond performance and impacts the financial feasibility of operating.
Blockchain Indexing is the process of querying historical data directly from the network and storing them in in-house provisioned infrastructure.
A typical and simplified Blockchain Indexing pipeline would have the following flow:
Use network API to fetch blocks and transactions.
Store blocks and transactions to internal storage infrastructure.
Use internally provisioned data query endpoints for products as opposed to querying directly from the network
This way, the Company has internal control over the Performance and Scalability of its historical data needs.
These read requests span across getTransaction, getBlock, getSignaturesForAddress.
While this is a solid metric of demand for BARA, the engineering team started experiencing bottlenecks around Bandwidth and Storage costs for requests directly funneled towards the Hypergrid infrastructure.
These bottlenecks were solved by decoupling typical archival read requests away from BARA's main Hypergrid infrastructure and into Sorada.
Sorada is BARA's Data Solution with the main objective of decoupling Archival Read Requests away from Network Write Requests.
By decoupling archival read requests away from the main BARA validator and into a separate data-optimized infrastructure, Bandwidth and Storage bottlenecks were mitigated.
Sorada enables SVM validators to release more compute resources to transaction processing, while optimizing read performance of archival requests by 30 – 40x.
According to BARA's internal Infrastructure monitoring, BARA experienced around 85% Archival Read Requests vs 15% Write Requests.