“`html
Building upon Protocol Update 001, we’re excited to present our methodology for blob scaling. The L1 functions as a strong basis for L2 systems to enhance Ethereum, and an essential element of secure L2 solutions is the data availability provided by the L1. Data availability guarantees that updates from L2 to the L1 can be verified by anyone. Blobs represent the unit of data availability in the current protocol, making it crucial to increase the blob count per block to facilitate a surge of L2 adoption for applications such as real-time payments, DeFi, social networking, gaming, and AI/interactive applications.
Our efforts are organized as a succession of gradual modifications to Ethereum’s blob framework. To expedite our scaling progress, we are transitioning from a “fork-centric” mindset to also integrate incremental optimizations in a non-disruptive manner as they become available. Consequently, we have numerous projects connected to both network updates and the intervals in between (“interfork”).
TL;DR
- Fusaka unveils PeerDAS, an innovative data framework enabling blob scaling beyond current throughput levels from 6 blobs/block to 48 blobs/block
- Blob Parameter Only (BPO) forks progressively elevate mainnet blob count, supported by additive peer-to-peer bandwidth advancements
- Innovative networking strategies anticipated for Glamsterdam will build upon the PeerDAS framework to enhance scaling further
- Mempool sharding upholds Ethereum’s principles as data continues to expand
- Exploration into the next generation of DAS reveals a development in secure DA scaling
PeerDAS in Fusaka
The initial milestone is the implementation of PeerDAS in the forthcoming Fusaka network update. PeerDAS introduces data availability sampling (DAS), wherein an individual node only acquires a portion of the blob data within a specified block. Alongside randomized sampling per node, the computational workload remains bounded, even as the overall blob count augments. As nodes are no longer required to download all the blobs in a block, we can increase the blob count without a corresponding rise in node demands.
Fusaka is anticipated later this year with implementations across all Ethereum clients. Comprehensive testing has been conducted on development networks (“devnets”) covering non-finality scenarios and adversarial “data withholding” situations. At this juncture in the R&D phase, we are intensifying the robustness of existing devnets and planning deployment to testnets and the mainnet. Barnabas Busa is spearheading efforts to guarantee a seamless transition through the final stages of the upgrade pipeline.
PeerDAS v1.x
We have two approaches to non-consensus modifications in our plan to progressively enhance blobs between the Fusaka and Glamsterdam upgrades: BPOs and bandwidth refinements. These are complementary as improved bandwidth utilization allows us to allocate resources towards increased throughput.
BPO
PeerDAS introduced in Fusaka establishes the groundwork for a theoretical amplification of 8x from the current throughput of Ethereum (i.e. ~64 KB/s to ~512 KB/s). Instead of instantly reaching this theoretical maximum at the time of Fusaka rollout, core developers have opted for a more gradual
“““html
increase via “blob parameter only” hard forks. This system enables core developers to program automated increases in blob capacity over time, maintaining a path of continuous growth. Once established, BPOs don’t necessitate any manual action to initiate. In between phases, we will observe the network and respond to scaling constraints that may only arise on the mainnet, clearing the path for the subsequent increase. Barnabas Busa alongside other members of the EF PandaOps team collaborate closely with the client teams to define the optimal schedule to achieve an 8x scaling from the present time.
Bandwidth optimizations
There is much we can execute to utilize bandwidth on the network more effectively. Raúl Kripalani along with Marco Munizaga are spearheading initiatives on this network engineering venture. A particularly encouraging optimization is the introduction of “cell-level messaging” which enables nodes to intelligently request components of the samples introduced in PeerDAS. This adjustment minimizes overlapping communication on the network, and the bandwidth efficiency can consequently be allocated to securely provision even more blob capacity. No consensus or execution protocol modifications are required to realize this milestone, thus they can be implemented interfork before Glamsterdam next year.
PeerDAS v2
This venture pertains to the next evolution of the PeerDAS design that provides even greater scale while leveraging the bandwidth savings achieved from pipelining introduced by EIP-7732 (projected for inclusion in Glamsterdam). Further enhancements to cell-level messaging and data reconstruction methods enable nodes to more flexibly sample distinct segments of blobs, allowing the core principle of DAS to be fully articulated. These improvements, along with the pipelining advantages that facilitate more effective utilization of the time between blocks, position us to scale beyond the constraints of current PeerDAS designs. There are numerous components at play, and precise figures must be calibrated based on both the performance of implementations and mainnet evaluations as the blob count is effectively scaled in a production context, but this effort should yield the ultimate multipliers on DA throughput prior to exploring alternative designs.
This set of updates will be incorporated in the Glamsterdam upgrade anticipated in the middle of 2026. Alex Stokes and Raúl Kripalani are managing the R&D efforts here to ensure we can continue scaling blob throughput.
Blobpool scaling
While the advantages of scaling are evident, we must carry this out while upholding Ethereum’s fundamental principles. One of these pertinent to blob scaling is censorship resistance. The mempool acts as a decentralized network for blob inclusion and directly facilitates censorship resistance amid a centralized builder network generating the majority of blocks on Ethereum. Although instances of censorship have improved over time, it is crucial for the scaling strategy to also ensure the blob mempool expands concurrently.
Csaba Kiraly is leading efforts here to sustain this vital resource. Current implementations support near-term blob throughput with vigorous research into optimal methods to scale the mempool as we reach higher levels enabled by Fusaka and beyond.
Future of DA
Beyond forthcoming versions of PeerDAS, we have numerous research paths to continue scaling DA while preserving the security attributes of Ethereum that distinguish it. Proposals generally fall under the label FullDAS with several variations under active exploration. A fundamental aspect of these proposals involves breakthroughs in peer-to-peer networking that accommodate a highly varied group of participants to shard an increasing quantity of samples while remaining fault-tolerant to adversarial entities. Activities such as Robust Distributed Arrays formalizes this concept. Additional considerations entail low-latency inclusion, censorship resistance, and advancements in the blob fee market to simplify the on-chain integration of blobs.
Research in this domain is guided by Francesco D’Amato and is quite active – feel free to reach out if you’d like to collaborate!
Source link
“`