Skip to main content

Relayer/TPU

TL;DR
  • What — Pre-leader Solana TPU traffic pushed to a gRPC Block Engine you operate.
  • Best for — Searchers building Order Flow Auctions (OFA), sandwich detection, or internal mempool aggregation.
  • Skip if — You consume confirmed blocks — use ShredStream for post-sequencing visibility.
  • Edge — Multi-validator coverage of pending tx (not just one TPU), real pre-leader visibility.

MEV mempool plus direct TPU submission. Two components, sold separately or together. The Relayer streams pre-sequencing TPU traffic to a gRPC endpoint you operate (Block Engine-style); the TPU adapter ships winning bundles direct to the validator's TPU — ~250ms priority window per leader, pay only if your bundle is profitable. Sourced from the Blockspace validator network.

Use this if…​

  • You're a searcher running arbitrage, sandwich detection, or signal-identification strategies.
  • You need pending-transaction visibility before the leader includes them in a slot.
  • You can host your own gRPC endpoint to receive transactions (Block Engine-style).

Architecture​

This is a push service, not a subscribe-and-consume one. Architecture is similar to Jito Relayer:

  1. Our validator's TPU receives a transaction.
  2. We forward it via gRPC to your endpoint.
  3. Your software (a Block Engine) processes / archives / acts on the transaction.

You operate the gRPC server. We're the client.

There's no off-the-shelf software that does this turn-key. You'll typically build your own or fork an open-source reference. Reference: jito-labs/block_engine_simple — the simplest example of what your software needs to look like.

What's different about Solana​

Solana doesn't have a global mempool the way Ethereum does. Transactions are sent directly to the current leader's TPU. The Relayer/TPU product gives you visibility into the pre-leader stream from Everstake validators, then your endpoint can join multiple validator streams to build a richer signal.

The signal is:

  • Richer than any single TPU — you see traffic addressed to Everstake validators across the network.
  • Incomplete by definition — transactions sent only to non-Everstake-network validators don't appear.
  • Coverage scales with how many validators opt in via Searcher Connection.

What customers build with this​

The TPU stream is raw orderflow data. Common downstream patterns:

  • Order Flow Auction (OFA) — share simulated tx signals with arbitrage searchers, collect bids in a short auction window, route winning bundles back to the network.
  • Sandwich and JIT detection — match incoming pending txs against liquidity-pool state to identify exploitable trades in flight.
  • Internal mempool aggregation — fan out the stream to multiple downstream consumers: your own validators, arb bots, internal monitoring.

You build the logic; we deliver the raw stream.

Relayer/TPU vs ShredStream​

Relayer/TPUShredStream
DirectionWe push to your endpointWe push UDP packets to your IP
SignalPending txs (before leader sequences them)Raw shreds (after sequencing, during broadcast)
AudienceSearchers identifying arb opportunitiesBots reading in-flight pool state
Latency vs slot boundaryPre-leaderPost-leader, pre-confirmation
Mature useSandwich, JIT, atomic arb on incoming txsPool-state tracking, snipers

You typically want both for a full searcher stack — Relayer/TPU to see what's coming, ShredStream to see what's actually being executed.

Next​