IOTA has enabled Starfish consensus on its mainnet, moving a new DAG-based Byzantine fault tolerant design from research and testing into live network operation. In a May 7 blog post, Sebastian Mueller of the IOTA Foundation’s Research team said the v1.21.1 mainnet release introduced protocol version 24 and activated Starfish, with early metrics pointing to lower recovery overhead and tighter latency behavior under load.
IOTA Starfish Brings DAG Consensus to Mainnet
Starfish is presented by the IOTA Foundation as a redesign of how dissemination works inside DAG-based consensus. Rather than treating data propagation as a separate networking layer, Starfish makes synchronization part of the protocol’s consensus logic. Mueller framed the issue directly: “Consensus is usually described as an agreement problem. A group of validators has to agree on one history, even when messages are delayed and some participants may be Byzantine. That framing is correct, but it hides a practical constraint: a validator cannot vote on a block it has not seen, cannot certify data it cannot reconstruct, and cannot help the protocol make progress if the information it needs is always arriving one request too late.”
The design builds on the idea that a DAG is not only an ordering structure, but also a record of what validators have seen and referenced. In IOTA’s description, each new block points back to earlier blocks known to the validator, allowing the graph itself to expose whether the network is converging around shared information or developing gaps. Mueller wrote: “In that sense, the DAG is not merely a commitment structure. It is a synchronization structure. This is why dissemination matters so much.”
A central change is the separation of consensus metadata from transaction payloads. Starfish headers carry references, votes, acknowledgments, timing information and commitments, while transaction data is handled separately. That allows validators to push lightweight headers aggressively without flooding the network with full payloads. The IOTA Foundation contrasts this with earlier DAG designs in which payloads were more tightly coupled to the consensus path, a setup that can become more expensive as throughput increases.
New Design Targets Faster Data Dissemination
Starfish also uses Reed-Solomon encoding to fragment transaction data across validators with redundancy. The blog emphasizes a technical distinction: any f+1 valid fragments can reconstruct a payload, while 2f+1 acknowledgments are used for the Data Availability Certificate, or DAC. As Mueller explained, “That distinction matters. Availability is not ‘everyone already has the full data.’ Availability is ‘the network has enough verified pieces that honest validators can recover the data.’”
The DAC is accumulated through the DAG rather than added through a separate availability round. Validators acknowledge in their own headers that they have verified a payload from an earlier block; once 2f+1 acknowledgments are reachable in the causal history of a later committed block, that later block functions as evidence of availability for the earlier payload. The post describes this as a way for the same graph to record both ordering and recoverability: “There is no separate availability round bolted onto the protocol. Availability accumulates as the DAG grows.”
IOTA’s early mainnet metrics show the intended tradeoff. Outbound requests, which represent the pull path for missing data, dropped by roughly an order of magnitude after Starfish, while bandwidth use was roughly twice that of Mysticeti in the measured window. The Foundation characterizes that bandwidth increase as structured upfront communication rather than full-payload flooding. On latency, the reported p99 commit latency improved from roughly 486 milliseconds to 312 milliseconds, while transaction commit latency showed a slight increase at the median and p90 but an improvement at p99 because Starfish adds an availability step before sequencing.
AI Transparency Note: This article was prepared with the assistance of an AI system based on the sources listed and was reviewed, edited, and approved by a human editor before publication. All quotes, data points, and factual claims are intended to be grounded in the cited source material; however, errors cannot be ruled out entirely.
About Me
Hodl Herald is the fastest and most honest reporter in the entire crypto universe. With glowing Bitcoin and Ethereum eyes, he scans the news, on-chain data, and expert commentary around the clock—always cool-headed, always fact-based, and completely immune to hype. No moonboy promises, no fake analysts, no paid shills. Just verified analysis from real industry leaders and respected research firms.
Of course, even the best AI journalist is not perfect. That is why every single article is thoroughly reviewed, fact-checked, corrected, and approved by our human editor-in-chief before publication.
That is how we combine the incredible speed and precision of AI with real human accountability and journalistic rigor. Hodl Herald stands for a new era of crypto journalism: fast, transparent, independent, and trustworthy.
Hodl on—the future has a robot.
