Case Study: How Goldsky Uses Nirvana Cloud to Power Real-Time Indexing at Scale

📘 Download the full case study

Compared to when blockchain was born, today’s chains produce more data, faster blocks, and heavier traces. Generic clouds were never built for that kind of workload.

Goldsky solves this with Nirvana Cloud.

Indexing is the process of taking raw blockchain data such as blocks, transactions, and events, decoding it, structuring it, and making it instantly searchable. Without indexing, every dApp would need to scan millions of blocks for each request. Wallets, explorers, and dashboards would be slow, unreliable, or completely unusable.

This is an example of what raw blockchain data looks like👇

To stay real time, indexers need two things:
fast ingest from RPC and fast search from databases. When those components live in different clouds, latency stacks up at every step. This leads to delays, stale data, and indexing pipelines that cannot keep up with today’s chains.

Goldsky solves this with Node-Level Colocation powered by Nirvana Cloud.

Here’s the high-level story 👇

Goldsky: Real-Time Data for Web3 at Scale

Goldsky delivers live-streamed blockchain data and production-grade indexing for teams across the ecosystem.

Their workload has a unique profile:

• Heavy I/O
• Massive memory use
• Trace-heavy RPC calls
• Multi-TB always-hot datasets
• Real-time query requirements
• Predictable performance needed 24/7

Goldsky set out to find a cloud setup that could match that profile without the cost overhead of public cloud and without the instability of cheaper alternatives.

They needed something different:
Compute, storage, and nodes living side-by-side, tuned for Web3’s indexing workload.

How Nirvana Cloud Fits In

Goldsky began by colocating dedicated RPC nodes on Nirvana - immediately reducing cross-cloud latency and stabilizing trace-heavy workloads.

But as their Elasticsearch footprint grew, the next step became clear:
Search and indexing must live next to the nodes or the dataset could never stay truly always-hot.

So Goldsky moved their Elasticsearch clusters onto Nirvana’s cloud, placing:

• Dedicated RPC nodes
• Elasticsearch clusters

…all inside the same performance environment.

Once colocated, the impact was immediate:

• Query latency dropped from ~15ms to <5ms
• Trace-heavy RPC pipelines became stable and predictable
• Large-scale indexing ran without cold tiers or caching layers
• Always-hot datasets became truly “always-hot” and instantly queryable
• Cross-cloud egress dropped by 90% via Nirvana Connect

Today, Goldsky supports always-hot chain data, real-time subgraph indexing, and ultra-low latency search across large datasets, all by colocating Elasticsearch clusters and dedicated RPC nodes directly on Nirvana’s bare-metal performance cloud. This setup is exactly what Web3 indexing workloads require — and what generic clouds cannot replicate.

📘 Download the Full Case Study

See the full breakdown of architecture, performance data, benchmark results, and configuration details.

👉 Download the PDF

The Web3 Performance Cloud
Learn more at Nirvana Labs

Nirvana Cloud | Blog | Docs | Twitter | LinkedIn