DoubleZero 백서

DoubleZero

원문

bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 3쪽 중 2쪽

have low value to a distributed system in isolation, but is highly valuable when operating alongside links from others. Indeed, the federated nature of the DoubleZero project is its core strength. Smart Contract-Defined Network The DoubleZero protocol organizes itself with verified data stored in onchain smart contracts. The distributed control plane uses this information to reach consensus on routing and prioritization decisions for the network. This provides a transparent and permissionless experience for both contributors and users. Contributors connect to the network and set a service level agreement enshrined in a smart contract, which includes endpoint locations, bandwidth, latency, and compliant MTU size. They send those details, along with the initial duration of the contribution, to the smart contract. Validators and other users can opt-in to predetermined services and routes, such as a specific L1’s traffic, state propagation, and validator set, or they can author their own contract to define a routing request. As inputs come in, the distributed controller uses smart contract data to identify the optimal configuration for the network. One key component of this configuration is the correct price, so that users pay for the cost of using the network in proportion to the request they impose on the system and providers earn incentives for contributing to the network in proportion to the value they generate in providing links. Ultimately, the project aspires to offer tiered pricing, where users can pay for priority access or base access to a given route to maximize network efficiency. The 10 DoubleZero network aims to offer a high degree of customization and control to its services and pass the surplus back to providers – a major advantage over the public internet, where traffic gets commingled regardless of priority Incentives and Alignment The open ethos of the DoubleZero project hinges on internal alignment and reciprocity by its network contributors. For the network to work, they must meet their service level agreements and have incentives to catch defection on the part of others. A system of checks and balances, based on the same cryptoeconomic values that secure proof-of-stake networks today, will underpin the DoubleZero system and ensure incentives are aligned for cooperation rather than defection. For instance, links that perform below their stated service level agreements will be penalized and ineligible for rewards (and eventually may be excluded altogether), while links that meet those obligations will be eligible for rewards. Unlike blockchain systems, physical networks lack the abstract verifiability of a block – where any node anywhere in the world can validate its correctness. Physical networks need physical connections (peers) to verify integrity, which today is only possible with redundant physical connections. DoubleZero will require new innovation for single point verification. Network Effects The DoubleZero project goes beyond a standard marketplace, where providers and users meet and negotiate bilaterally. It is a true network – each link is more valuable as part of the whole than it would be in isolation. Consider a link provider that has a few disconnected links, which represent underutilized capacity on their internal networks. While the entity could sell services over those routes piecemeal, each route would require substantial overhead in finding buyers and negotiating contracts. More importantly, each link has limited value, as it serves only one particular route. But the provider who offers those links to DoubleZero benefits in two ways. First, the process of adding contributions to the platform is easy and standardized (all one needs to do is install and configure a few switches), requiring little overhead. Second and more critically, the link grows in value by its association with neighboring links. A disconnected link can potentially become a highly connected link on the DoubleZero network, making it substantially more valuable than it would be on its own. Network effects are present in many other markets. One example: airlines. A single airline route between two cities is rarely offered in isolation, as there is not likely much customer demand. Instead, they are bundled into a carrier’s larger network and interact with other routes through hubs. These bundles are substantially more valuable; they become even more valuable when the carrier joins an airline alliance and implicitly ties their routes with routes in other parts of the world. DoubleZero is the platform that takes individual routes and turns them into a thriving cross-connected network. 11 The DoubleZero network can even benefit from redundant links, short links, or low-performance links. While the network is high-performance and expansive holistically, these types of contributions add redundancy, capacity, and interconnectedness. Thus, the DoubleZero project can allow providers to monetize capacity that otherwise would have almost no value in isolation. Taken together, providers who make unused capacity on their network available to DoubleZero face lower transaction frictions and generate more user surplus from the network economics, especially when compared to selling that capacity in isolation. In turn, the DoubleZero network passes that surplus onto the providers. Blockchain Use Cases Blockchains, as the highest-profile distributed system archetype, holistically represent the strongest use case for the DoubleZero network. A blockchain's dual objectives of being public and performant often work against one another in the status quo. Across contexts as diverse as Layer 1 blockchains, RPC nodes, MEV systems, and Layer 2 chains, the DoubleZero network can push towards attaining both. Figure 6. A conceptual map of the DoubleZero Network 12 Layer 1 Blockchains A Layer 1 blockchain is the distributed system that stands to gain the most from the DoubleZero network. In the status quo, each validator faces a firehose of inbound transactions (including spam and duplicates). Consensus is reached slowly as validators propose blocks and vote over public internet paths prone to high jitter and inconsistent routing. In tandem, users suffer from poor experiences, whether it be competing with a flood of spam or waiting for delayed finality. Even today’s most performant blockchains fall short of providing both the web2-style experiences users expect and the professional central venue trader experience. The industry has developed some solutions, but each comes with disadvantages. Mempools serve as a central holding pen for transactions (to lower the ingestion burden on individual validators), but this adds latency to the system and broadcasts transaction intentions well before their inclusions. Quality-of-service solutions have been tested on blockchains like Solana wherein validators accept transactions predominantly from those who have staked capital, but this excludes entities who do not have sufficient stake from full participation, does not guarantee economic efficiency, and adds additional latency from the proxy step. Validators can invest more in hardware to keep pace, but this of course inhibits the development of a diverse and decentralized validator set and does not address aggregate network performance. The DoubleZero network offers a more robust solution in two ways. First, it provides the filtering technology to filter inbound traffic that comes over the public internet. This step utilizes scalable hardware running open-source code, mitigates distributed denial-of-service attacks, verifies signatures, and removes duplicate transactions. This separation of filtration, deduplication, and signature verification from block production, execution, and consensus frees up substantial resources to focus on block production. This is cooperative infrastructure sharing, where the system in the aggregate can share resources rather than every validator individually needing them. 5 Second, it routes the filtered traffic – whether user transactions, blocks, or consensus votes – to the validator set over its dedicated bandwidth links with minimal latency and jitter. This ensures that blocks are built, shared, and ratified quickly. Under this networking paradigm, high-performance validator clients or consensus algorithms can come much closer to their theoretical maximums. Indeed, this can be optimized further as the DoubleZero network is expected to support multicast functionality, which allows for better propagation of state transitions, blocks, and shreds. There is one final advantage of using the DoubleZero network for Layer 1 blockchain communication. Packets traversing over the network are expected to be traceable so that users can verify their inclusion or exclusion from intra-validator communication. For systems that are sensitive to concerns of censorship, this is a beneficial by-product of transparency in traffic routing. 5 There are many other successful examples of cooperative infrastructure sharing, including mobile networking, TowerCos, or subsea fiber optic cable system consortia. 13 RPC Nodes Remote Procedure Call (RPC) nodes can similarly benefit from the DoubleZero protocol. As the primary interface between users and the chain, they face especially severe filtration and communication challenges. RPC nodes comprise the infrastructure that mediate user interactions on a blockchain. The primary role of an RPC node is to receive inbound transactions from users and to forward those transactions to either the leader or the chain's mempool. The secondary role is to provide an updated view of the chain to downstream applications, such as DeFi applications, block explorers, or gas estimators that help users form efficient transactions. RPC nodes are effectively non-voting validators – they track state transitions on a blockchain, but do not participate in consensus. There are three challenges that RPC nodes face that could be mitigated by the DoubleZero protocol. First, RPC nodes are directly exposed to traffic surges from users, particularly around broad community events like airdrops or NFT mints. Because they specifically receive the transaction flow via the public internet, they are especially vulnerable to distributed denial-of-service attacks. Second, the role of the RPC node is to forward user transactions to the block leader or the mempool, depending on the blockchain. Transaction deliverability matters immensely for an RPC, especially since the highest-value transactions (e.g. arbitrages) tend to be both competitive and time-sensitive. Resilient and low-latency connections from the RPC node to the block leader or mempool can assist RPC nodes in these responsibilities. Third, RPC nodes return information from the blockchain to downstream users, wallets, and dApps. They are primary providers of data about the state of a network, providing everything from wallet-balance information to available liquidity for a given trading pair on a DEX. In order to do this, they need to ingest, validate, and index changes as close to real time as possible. Resilient and low-latency connections back from the validators to the RPC nodes can ensure that RPCs have up-to-date views of the blockchain. The DoubleZero protocol can make the RPC ecosystem more competitive. RPCs already vie aggressively amongst themselves on timeliness of information both to and from the chain. The DoubleZero network opens an avenue for them to improve service yet further. Maximal Extractable Value The DoubleZero network can support a more fair and open maximal extractable value (MEV) ecosystem. These systems are extremely latency-sensitive, such that even small improvements in latency and deliverability can deliver additional value. MEV refers to the value that block producers can take by ordering transactions in a block – by strategically adding a transaction to close a price dislocation caused by a large trade. In practice, mature MEV systems typically situate themselves as adjacent to the validators. They receive proposals from external traders (known as “searchers”), identify a profitable ordering of transactions, and forward the ordering to the block leader along with some additional payment. 14 Reliability and latency in communication matter on two dimensions. First, MEV systems benefit from a more real-time view of the state of a network. With fresher data, they can construct more profitable transaction ordering, allowing them to craft more efficient blocks. Second, even holding the opportunity set fixed, MEV systems that can relay proposals to validators quickly can use the extra time to find better orderings. Block construction is a complex combinatorial problem that requires approximation and simulation algorithms to find local optima – any extra time allows those algorithms to run longer. Put another way, MEV systems benefit from faster retrieval of data (read operations) and faster delivery of blocks or bundles to the current leader (write operations). Increasing bandwidth and reducing latency improve both of these factors. Layer 2 Blockchains Layer 2 blockchains that use central sequencers or a distributed set can utilize the DoubleZero network too. In a typical Layer 2, a single sequencer posts inbound transactions to a Layer 1 chain; a set of Layer 2 validators then use those to update the chain state. There are standard ways that a Layer 2 chain can utilize the DoubleZero protocol, such as filtering inbound transactions to the sequencer, posting to or retrieving from the Layer 1 without latency, or verifying updated state among the validators quickly. There are two more critical ways that Layer 2 chains can benefit from the DoubleZero network. First, most Layer 2 chains aspire to support multiple sequencers (posting transactions either in rotation or in tandem) to reduce centralization, censorship, and liveness risks. This architecture shift requires substantial coordination between the sequencers to prevent wasted computation for both the Layer 1 chain and the Layer 2 validator set. The DoubleZero network can smooth that communication between sequencers. Second, many Layer 2 chains aspire to scale using data availability layers, which store transaction data, smart contract data, and even off-chain data on separate systems that can deliver information on demand. But the efficacy of this modular architecture is limited by the bandwidth connections between the modules, and low-bandwidth connections can lead to new bottlenecks in the underlying chain. Again, the DoubleZero network can ensure that the data availability layers keep pace with Layer 2 chains. Other Blockchain Contexts The DoubleZero network offers advantages in other blockchain contexts: state synchronization, indexers, multiple leaders, and network extensions, among others. One challenge with high-throughput blockchain systems is onboarding new validator nodes. These nodes must synchronize to the current state by replaying transactions from the chain start, as outlined by MegaETH [2024]. For high-throughput blockchains, this can impose too much pressure on existing resources like bandwidth. The new validator node cannot simply ingest transactions at the current speed of the blockchain but must ingest substantially faster in order to catch up to the 15 current state. The high-bandwidth connections of the DoubleZero network can be useful in overcoming these constraints and making it easier for new validators to join. By the same principle, it can also help validators that fall behind come back online. As blockchains scale, supporting infrastructure like indexers will become more important. Like RPC nodes, indexers need to receive updated state quickly so that their end users can form quick and reactive decisions. State propagation through DoubleZero reduces delays to indexing fresh data, and can help keep distributed indexer instances current. Many Layer 1 blockchains aspire to have multiple leaders simultaneously produce blocks in different parts of the world. Without tight coordination between leaders, though, this can quickly lead to redundant or outdated transactions and wasted compute. Like multiple sequencers on Layer 2 chains, multiple leaders on Layer 1 chains need rapid global state synchronization; and the DoubleZero network can facilitate this better than the public internet. Lastly, there are a whole class of blockchain extension services that do not fit nicely into the “Layer 1” or “Layer 2” buckets. These uncategorized services are sometimes referred to as “network extensions.” Nearly every network today has the concept of a compute operation that can be run on only a subset of a network’s validators, because it requires specialized hardware or software. ZK proovers, GPU workloads, and asynchronous execution platforms all can benefit from the increased bandwidth and reduced latency links DoubleZero provides. Other Use Cases The DoubleZero network has uses beyond blockchain technology. Any setting in which distributed systems have to communicate quickly, precisely, and voluminously can benefit from its structure. This section lays out four additional examples. Content Delivery Networks At a high level, content delivery networks (CDNs) look similar to the DoubleZero network. CDNs make content delivery more efficient for users by filtering public requests and routing them efficiently through their network. This, coupled with local caches that hold content, allows for users to enjoy reliable and fast experiences for a given web application. A new CDN can rely upon the DoubleZero network to bootstrap itself more quickly and focus on its core competencies. CDNs excel at forecasting demand and provisioning content accordingly; but in the status quo, they must also filter traffic and establish networks with low latencies and high bandwidth to offer an end product. 6 A new CDN built atop the DoubleZero network could return to those core competencies only, and leave the filtering and routing to the protocol. 6 The incumbents, like Akamai and Cloudflare, mostly establish long-term leases from network service providers to structure their network. 16 Moreover, the DoubleZero network can boost the operations of existing CDNs. One major advantage of the DoubleZero protocol is its flexibility: Users do not need to contract for expansive and long-term leases, but can customize their usage for specific routes and time intervals. Thus, CDNs that maintain their own network but lack coverage in certain regions or need more bandwidth across central corridors can use DoubleZero accordingly, offering better experiences to users and relieving strain on their networks alike. This could be especially powerful as multicast functionality comes to the network and allows for more efficient delivery of content. Online Gaming Latency makes and breaks online games. Even mild lags ruin the user experience and restrict players to simple turn-based games. Traditional game developers have used two models: peer-to-peer models or central servers. The DoubleZero network can augment both. Some games use peer-to-peer models, whereby the end players connect directly over the public internet. 7 These are relatively efficient, particularly when there are few players, but these models are directly constrained by the limitations of public internet. Latency, constrained bandwidth, and jitter can lead to frustrated users. The DoubleZero network, with its dedicated links and its flexible usage model, can deliver stability to the connections and make the games more enjoyable. Other games use the model of a central server to which all players connect. These models are particularly used when there are many players. Historically, these games have been built atop cloud providers, which have their own set of advantages and disadvantages. Cloud providers offer stable performance for players close to the edges of their network, but p
프로젝트로 돌아가기