DeServe.network Global Paseo Archive RPC Deployment - Proposal #1
ℹ️ Please view the counterpart proposals:
Summary
This proposal is for the global deployment of a progressively decentralized archive RPC network for the Paseo testnet and its system parachains. Please view the detailed forum post for the complete background of the initial deployment for Polkadot Asset Hub and Coretime.
DeServe is:
- Global RPC infrastructure - 14 locations, widest coverage
- Geo-steered load balancing
- Lowest latency among all major Polkadot RPC providers
- By a Polkadot-native infrastructure & software builder, ex-IBP Rank-6 member
- Path to on-chain, general-purpose infrastructure (DePIN)
- Launched on March 31st, 2026 as a single-chain deployment, >53M req/24h by day 10
- Testnet live on Paseo, para id 5150 (telemetry), proof of on-chain commitment, currently running a shell runtime with business logic upgrade planned
- 80% cheaper than IBP
- Webpage live at deserve.network
- Developer documentation: docs.deserve.network
This proposal requests $11,719.00 as the first payment, half of the first three months, to bootstrap the global rollout of all Paseo services. Subsequent payments will be requested via 3-monthly retroactive proposals, alongside full transparency reports covering expenses, performance, and request analytics.
Initial Traction & Updates
The following X posts document DeServe's first days of deployment:
- Launch announcement for Polkadot Asset Hub (post)
- >1.7M requests on day 1, Ethereum-compatible RPC deployment (post)
- >45M requests/24h, deserve.network launch (post)
- >50M requests/24h, docs.deserve.network launch (post)
- >53M requests/24h, Polkadot Coretime rollout and public dashboard launch (post)
Over the first two weeks of deployment, DeServe has served over 500 million requests, and is currently serving >36M requests/day.
Live request analytics are available at the public dashboard.
Deployment & Coverage
Complete Paseo rollout under this proposal
- Paseo Relay Chain
- Asset Hub
- Bridge Hub
- Collectives
- Coretime
- People
| Chain | Archive RPC | ETH RPC | Status |
|---|---|---|---|
| Paseo Relay Chain | ✅ | - | To be deployed |
| Asset Hub | ✅ | ✅ | To be deployed |
| Coretime | ✅ | - | To be deployed |
| Bridge Hub | ✅ | - | To be deployed |
| Collectives | ✅ | - | To be deployed |
| People | ✅ | - | To be deployed |
Infrastructure
- 14 locations across 6 continents, 7 providers including Helikon
- Atlanta, GA, US
- Gravelines, FR
- İstanbul, TR
- Johannesburg, ZA
- Limburg, DE
- London, UK
- Miami, FL, US
- Montreal, CA
- Mumbai, IN
- Seattle, WA, US
- Sao Paulo, BR
- Singapore, SG
- Sydney, AU
- Warsaw, PL
- 1G–10G uplink bandwidth
- Unmetered traffic on most nodes
- 950GB–14.72TB NVMe storage per node
- ECC RAM
Monitoring and High Availability
- Geo-steered load balancing: requests routed to the nearest node
- 15-second health checks: unreachable nodes removed from the pool
- Quick recovery from low-cost backup storage preferred over 2N redundancy for cost-efficiency
- Prometheus & Grafana monitoring and alerting across all nodes 24/7
- Monitoring network (in development): latency checks across the globe
- 99.9%+ effective availability through rapid failover
- Open-source DNS transition planned (see Roadmap)
RPC configuration
- 100 req/s without API key
- 2MB max request size / 7MB max response size
- 256 RPC subscriptions per connection
Performance
DeServe delivers the lowest latency among all major Polkadot RPC providers, verified via Compare Nodes, a global RPC performance inspector.
| Comparison | Continents | Regions |
|---|---|---|
| DeServe vs IBP | 5/6 | 18-20/26 |
| DeServe vs OnFinality | 6/6 | 25/26 |
| DeServe vs Dwellir | 6/6 | 24/26 |
| DeServe vs LuckyFriday | 6/6 | 24/26 |
Full benchmark runs:
- DeServe vs IBP: Run 1, Run 2, Run 3
- DeServe vs OnFinality: Run 1
- DeServe vs LuckyFriday: Run 1
- DeServe vs Dwellir: Run 1
Cost Comparison
IBP is currently the only other provider offering a global geo-steered RPC service, making it the most appropriate reference for cost comparison.
Note that IBP operates on a 2N redundancy model, while DeServe operates on a single-instance model with rapid failover through backups, as detailed in the Monitoring & High Availability section.
Given the current cost constraints of the Polkadot ecosystem, we find that single instance per location with backup mechanisms in place should be sufficient under a highly responsive load-balancing system. DeServe’s current setup monitors endpoints every 15 seconds, immediately removing any unreachable endpoint from the pool.
IBP costs for Paseo are taken from the same billing view as the Polkadot and Kusama figures.
| IBP | DeServe | |
|---|---|---|
| Paseo Relay Chain | $17,428.10 | $3,486.00 |
| Asset Hub | $5,101.06 | $1,020.00 |
| Ethereum RPC | $538.30 | $107.00 |
| Bridge Hub | $4,002.82 | $800.00 |
| Collectives | $4,002.82 | $800.00 |
| Coretime | $4,002.82 | $800.00 |
| People | $4,002.82 | $800.00 |
| TOTAL | $39,078.74 | $7,813.00 |
| vs. IBP | - | 80% cheaper |
Our cost model is illustrated in detail in the initial forum post. DeServe also provides GeoDNS services for free as part of the package, whereas IBP charges $1,200.00/month for this service. DeServe also comes without curator payments, which add further overhead to bounty-based programs. IBP curator payments add approximately $3,400.00/month in additional overhead (reference).
Operation & Payment Details
Payment model
- First payment $11,719.00: half of the first three months ($7,813.00 × 3 × 0.5), requested upfront to bootstrap the global rollout
- Second payment $11,720.00: remaining half of the first three months, requested at the end of month 3 alongside the first transparency and performance report
- Subsequent payments $23,439.00/quarter: 3-monthly retroactive proposals alongside full transparency reports
Payment terms
- Stable or native coin (USDT, USDC, or DOT): first proposal submitted in USDT given current DOT valuations
- No lock-in: the treasury can cancel anytime
- Via regular spends: no long-term bounty top-ups or commitment
- Monthly transparency reports: expenses, performance, and request analytics
- Public monitors: service health and performance, live at public dashboard
- 99.9%+ uptime SLA: basis for payment validation
Labour costs
Labour costs for the current alpha phase are excluded from this proposal and covered by Helikon. As DeServe matures, operator labour costs will be standardized through protocol governance.
Roadmap
First 3 Months
- All Paseo services live within week 2 of proposal approval: Paseo Relay Chain, Asset Hub, Bridge Hub, Collectives, Coretime, People
- Monitoring network: real-time latency and health checks across all regions
- DeServe website: live node map, performance data, endpoints, and developer documentation (live at deserve.network)
- Open-source DNS: replacing Cloudflare for geo-steered load balancing, integrated with the monitoring network
- Polkadot-native provider onboarding: begin replacing cloud providers with native Polkadot infrastructure providers, onboarding criteria to be published
- Governance alpha phase: managed and governed by Helikon
- DeServe Unpaper: protocol design and vision document
First 6 Months
- Complete Polkadot-native operator onboarding
- Governance beta phase: network committee of ≥3 native operators
- On-chain service provision: on-chain payments, proofs, and governance, built on Polkadot
Submerge Commitment
Submerge is a data and compliance platform for Polkadot SDK chains, currently in development by Helikon. Submerge has received treasury funding and is behind schedule. Its two main components, Crystal (a chain indexer), Mycelium (a cross-chain indexer), along with their APIs for all supported chains will be delivered before any further on-chain submission for DeServe.
About Helikon
Helikon is a Polkadot-native infrastructure and software development collective based in İstanbul. A regular contributor to the Polkadot ecosystem at both the development and governance levels since late 2020:
- 5 active Kusama & Polkadot validators, former 1KV & Decentralized Nodes member
- Rank-6 ex-IBP member: infrastructure services for 26 blockchains
- Two-time Decentralized Voices delegate, once as founder of Permanence DAO
- Recently saved and revived Multix (post 1, post 2, post 3), a >$500K investment funded through Polkadot treasury proposal #236 and referendum #120, and child bounties 36_10 and 36_3054.
- Maintaining SubVT (iOS & Telegram Kusama/Polkadot), Chain Console, Chainviz/alpha, followthedot
- Built dv.report - analysis dashboard for the Decentralized Voices program cohorts 4 and 5
- Developing Submerge - data & compliance platform for Polkadot SDK chains
Contact
Comments (2)
Requested
Proposal Failed
Please view our responses to Rust Syndicate's questions on the Kusama proposal here.
I’m voting Nay, but I do think this is pointing toward a good idea that could be valuable with some refinement.
At the moment, there are two main funding sources for RPC services: the IBP and the Public RPC Bounty.
The IBP is relatively cost-heavy because each member needs to be paid for a baseline level of service. Its major advantage, however, is that it provides a single GeoDNS endpoint, making it convenient for users to connect to the RPC node closest to them.
The Public RPC Bounty, on the other hand, uses a more traditional procurement process: scope, tender, and bid. This generally provides better value for the community, but it does not currently benefit from a single GeoDNS setup.
DeServe is a good idea because it extracts the GeoDNS concept from the IBP and proposes delivering it at a significantly lower cost. My concern is that the resources at each geographic location may not have the same benefits of being self-owned and operated, as is the case with the IBP or, in some cases, the Public RPC providers.
Have you considered a hybrid approach?
One possible model would be to have one or two central DNS providers, ideally using different providers for redundancy, while allowing bids to occur per geographic location. For example, providers could bid to serve regions such as US West, US East, Europe, Asia, South America, and so on.
The tender evaluation could then award additional points for factors such as self-owned infrastructure, bare-metal deployments, operational maturity, and whether the provider has dedicated staff rather than being a one-person operation.
I think this could preserve the strongest part of the DeServe proposal, the shared GeoDNS layer, while still using competitive procurement to ensure good value, regional coverage, and higher-quality infrastructure.
Hi @Paradox,
Thanks for the feedback. DeServe's proposal already is to immediately start replacing cloud providers with Polkadot-native operators upon approval. Please see 'First 3 Months' under 'Roadmap'. Tender approach for these specific locations is an improvement on top that I'd definitely be willing to apply. An on-chain implementation would be preferable, and it would improve all infrastructure services in the ecosystem by combining the best approaches:
And just to note, one function of the DeServe proposals is to form an initial cost baseline for the listed RPC services.
Best regards.