Performance

We are committed to both cost optimization and performance enhancement. By relocating our database from commercial cloud environments, which often come with multiple layers of networks, we inherently improve performance. However, our strategy extends beyond this.

Our ArchivalRPC server is designed to be significantly more efficient than traditional heavy validators that are typically used for RPC services. This allows it to process more requests per second (RPS) with reduced demands on CPU and RAM.

We are continually developing our ArchivalRPC server with a clear objective to reduce all RPC request response times to below 20 milliseconds. We are approaching this target, demonstrating substantial progress in our performance goals.

Whats in the box

Each SVM archival setup is deployed on dedicated bare-metal machines. We work with multiple bare-metal providers and can offer numerous locations around the globe, always preferring to work with those that do not impose bandwidth limitations, so the only constraint is your dedicated hardware.

All software is deployed to our internal Kubernetes clusters; upon request, we can start a dedicated Kubernetes cluster and give you full access to it. From hardware to the data we write, everything can be customized. If you don't care about voting transactions, we can skip them along with many other filters to optimize storage and costs.

ArchivalRPC is just one small component in the entire archival process; the biggest challenge is the ingestion process, keeping track of the entire chain, not missing any blocks, having incremental backups, and the ability to quickly bootstrap new instances or perform healing operations for existing ones. Check out the full list of containers that are running for just a single network 👉