Skip to content
snapshot

Query Console

Compose p2p_query / p2p_join / p2p_share over data in S3/ADLS/GCS. Several hosts run it redundantly; the first correct result that reaches quorum wins.

Compose a query and watch the grid run it. Instead of one server, several machines run your SQL at the same time and must agree on the answer before it is accepted.

Impact: You get a fast result that has already been cross-checked — so a single broken or dishonest machine cannot hand you a wrong answer.

Compose
Pick a grid function and write SQL against object-store paths.
Execution policy
Every p2p_query override — scope, routing, verification, trust, and economics.

Scope

Routing

Verification

3

Matching result hashes required before a result is accepted.

4

Hosts the query races on redundantly.

Trust & attestation

Effective policy≥ L1 attestationtrust ≥ 0.85paid · stake countsauto (from data class)6 eligible6 hosts eligible — quorum capped at 5.
auto · L1
auto · 0.850.85

Excludes hosts with no bonded stake.

Economics

Requester locks up to this escrow; the unused remainder is refunded.

Candidate set and result preview are seeded from the real grid snapshot (honest workers by trust; the result the grid actually computed).

Live run
Hedged execution across 4 hosts · verify Quorum · quorum 3
OfferBiddingDispatchExecutingCommitVerifySettle
Configure the policy and dispatch to watch the race.
Translates to
The QUIC wire messages this dispatch emits, in order.
  1. Offer{ job_id: "job_pending", query_hash: "b3:f4ce16d8f4c…", data_class: Internal, nonce: 0 }
  2. Bid{ decision: Accept, eta_ms: 14, price: 7, attestation: L2 }
  3. Dispatch{ sql: <…56 chars>, credential: ScopedCredential, memory_limit_bytes: 4 GiB, threads: 8, verify_mode: Quorum }
  4. Commit{ result_hash: "6ac9cf852bcc8d…", row_count: 3, latency_ms: 14 }
  5. Receipt{ verdict: Correct, sig: "ed25519:6ac9cf85…" }

Offer → Bid → Dispatch → Commit → Verify → Settle

commit-first