ERPC powers up Solana Shredstream shared endpoints in three high-demand cities, strengthening low latency and stability
ERPC powers up Solana Shredstream shared endpoints in three high-demand cities, strengthening low latency and stability
ELSOUL LABO B.V. (Head office: Amsterdam, the Netherlands; CEO: Fumitake Kawasaki) and Validators DAO announce a power-up maintenance for the Solana Shredstream shared endpoints in the highest-demand regions of Frankfurt, Amsterdam, and New York, delivering further improvements in latency and stability.
To match the increase in usage and demand, we applied multiple nodes with our highest-ever clock configurations and expanded headroom across compute and distribution. There are no changes to existing connection methods, pricing, specifications, rate limits, or authentication.
We sincerely appreciate your continued support.
Why we strengthened the shared endpoints and what changed
Shredstream is the foundation that receives order-book updates and in-block events via the shortest path and accelerates your application’s decision and execution.
Over the past few months, concurrent connections and stream processing on the shared endpoints have continued to rise, so we needed to further increase peak-time headroom and stability.
We added multiple server configurations with market-leading clock speeds and focused tuning on relieving memory pressure, expanding page cache and ring buffer headroom, and suppressing queue congestion during spikes.
By reducing choke points across both the processing and delivery paths, we aim for behavior that resists slowdowns and drops even during peak periods.
Why we focused on Frankfurt, Amsterdam, and New York
On Solana, the leader validator rotates globally on a short schedule. Validators are heavily clustered in Frankfurt, Amsterdam, and New York, with approximately 50% concentrated across these three metro areas.
By covering these three locations, your application can run in close proximity to the leader for about 50% of the time, minimizing external network round-trips, which are typically the most time-consuming part of computing. Cutting distance directly improves speed and consistency.
Shared power-up and dedicated options
The shared endpoints are designed to deliver a high-performance data stream to many users with strong cost efficiency, and this power-up improves peak resilience and the ability to hold low latency.
At the same time, because shared use is assumed, workloads that require extremely strict latency ceilings or always-on “fastest” consistency are better served by dedicated configurations.
Dedicated Shredstream pricing depends on the CPU and server capability, and the most popular choice is Metal Ryzen. We provide configurations suited to workloads that demand constant top speed.
Your application’s location determines latency
Speed is not decided by server specs alone. Where your application lives matters just as much.
As an extreme example, continuously monitoring what happens in Frankfurt from Tokyo is disadvantageous. Round-trip network latency alone accumulates delay and keeps you behind.
Preparing resources in each region and either completing ingestion and processing locally or bypassing to the next region via the shortest route raises both coverage and responsiveness.
VPS on the same network
Our VPS instances are deployed on the same network as the Solana dedicated endpoints in each region, removing external network hops and enabling the shortest round-trip.
They can be deployed quickly and in small units per region, and even distributing 1–2-core workers improves effective latency and resilience against drops.
About SUPER EPYC VPS coming in September
Starting in September in Frankfurt, our most popular region, we plan to launch SUPER EPYC VPS, built on a data-center CPU that achieves a market-leading 5.7 GHz clockspeed.
Because adopting the latest-generation CPU for a VPS product is uncommon, supply will be limited. It is a strong option when you need the fastest VPS.
For the highest quality and speed, choose bare metal
Whereas a VPS divides a physical server, a bare-metal server lets you exclusively use the entire machine: CPU, memory, disks, and even network bandwidth.
This makes it easier to maintain high stability and speed during peak times and is well-suited to Solana applications that require consistently low latency.
For Solana workloads, the Ryzen lineup is especially popular, reaching a 5.7 GHz top clockspeed despite being consumer-class. EPYC is designed to reduce virtualization overhead, while Ryzen is designed to maximize single-machine performance without virtualization. Choose according to your use case.
Challenges ERPC solves
- Transaction failures and latency fluctuations that tend to occur in general RPC environments
- Performance throttling by many infrastructure providers
- The large impact that network distance has on communication quality
- The difficulty for small projects to access high-quality infrastructure
For product details, free trials, onboarding, dedicated configurations and inventory inquiries, and waitlist registration, please contact us via the Validators DAO official Discord.
- ERPC official website: https://erpc.global/en
- Validators DAO official Discord: https://discord.gg/C7ZQSrCkYR
We will continue our research and development, stabilize inventory, and expand our lineup to deliver value to more projects.
Thank you for your continued support.
News
