We’ve been at the forefront of evolving PBS, pioneering the emergence of new functionalities like preconfirmations and inclusion lists. A core tenet to making this a reality and not introducing additional sidecar software is to add new functionalities to relays, with the goal being to create entirely new verticals that bring additional vectors of competition and specialization for relays beyond that of latency. Not only can added functionality help relays find sustainable income models but it can also increase the promise and potential of PBS while achieving goals aligned with the Ethereum ethos.
Our new functionality provides new capabilities that enable relays to read validator preferences and enforce those preferences through block filtering while maintaining their core block routing performance. This is applicable not only to leverage preconfirmations, but also to segment the market based on validator preferences for OFAC compliance and non OFAC compliance, among other sought-after use cases.
We created a grant request to PBS Guild to build this new relay functionality which wasn’t approved, citing our funding sources. However, we knew this was important for the ecosystem and core to increasing PBS functionalities without introducing additional validator sidecars so we took it in our own hands to develop this capability in the open and make it available to all existing and potential new relays.
Relays can now enforce validator preferences at scale by reading from a smart contract and enabling validators to conform to the rules of new applications by exposing a generalized filtering interface. This filtering is critical because PBS operates outside Ethereum's core protocol - builders could potentially make commitments under one identity while building blocks under another. For preconfirmations, relays can now verify that blocks come from registered builders for commitment enforceability, mitigating a key sybil resistance problem arising from a lack of builder identity in protocol. By handling filtering at the relay layer, we avoid adding complexity to the PBS pipeline while ensuring commitments remain credible.
As PBS continues to evolve, this filtering capability provides the foundation for relays to integrate with new applications while maintaining their essential performance characteristics, and we’re excited to bring it forward in open source.
As a result of implementing these filtering capabilities, we encountered an impact to performance. When we first implemented the required checks on Holesky, each block submission required two Redis lookups, one to check validator registration and another to verify builder registration.
These checks were adding up to 1ms latency at times per block submission. Although in some circumstances, a millisecond doesn’t sound like much, this was potentially problematic for mainnet deployment.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXcTkXWWjGYpQTpAQPqXmFKKHjtkE2FCyxM-mgyR2qUSA5GnurbHtPBj5fHcHAT0y0Y_HXE7L0vdx8AJnQgADeVFXPesdrBFo9j-JtfOA4jRwMzvbuxqt_9Eb-PKF4a4yaD7XeKKGAOkCI4hJGV-SAPgHKIZ?key=2kqUibyyYJeJCGwklAAJHw
We resolved this by moving to local caching for the checks instead of Redis queries. This resulted in a 1000x performance improvement, reducing the overhead to just a few nanoseconds.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXdopSQw69UQeL3fVuYxIzvSpGYay_MuDveHx7BK4ksEEXO0GFH1p47zvoOYMQ5ZrSwbJ27DBniPjPcZZPO_mO80Bznvr9R1T00mvNPXwnjPxPssWRKmcDptOdqH-g2Wf6ttZitMDbXAvonipySlfMayc4Sa?key=2kqUibyyYJeJCGwklAAJHw
Relays have now integrated this optimized filtering and are delivering protocol compliant blocks on Holesky testnet. The implementation ensures validators only receive blocks from registered mev-commit builders, with virtually zero performance impact.
Optimizations at this level have massive implications for network dynamics and fairness. By reducing filtering overhead to near-zero, we've removed a major barrier to relay adoption of mev-commit’s credible commitments. More importantly, this opens up avenues for relays to support a variety of validator preferences beyond preconfirmations: