# Evolution of the Internet and Blockchain: The Looming Quest for Scalability

## Key Takeaways

The history of the internet is intrinsically linked with speed. As the speed and capability of the internet increased, it enabled the development of dynamic Web 2.0 applications, creating the modern internet ecosystem. Similarly, blockchain, to fully address applications like the metaverse, AR/VR, and more, must achieve significantly greater scalability.

Recent blockchain projects like Solana, Sui, MegaETH, and Fuel have all focused on improving scalability through parallel processing. However, while parallel processing aids scalability under typical conditions, it falters during high-demand events involving similar states, such as popular NFT minting events.

Somnia, on the other hand, distinguishes itself by maximizing sequential processing of EVM transactions within a single core. It converts EVM bytecode into native code executable by the CPU, leveraging hardware-level parallelism within a single processor core.

## Background: The Pursuit of Parallelism

### The Internet’s Evolution with Speed

Historically, advancements in internet applications have paralleled increases in internet speed. The TCP/IP standard and Tim Berners-Lee’s World Wide Web laid the groundwork for the internet. Early on, data transmission was limited to text-based emails and simple web browsing (below 56 kbps).

Mid-1990s saw a shift from dial-up to DSL and cable modem connections, allowing for audio and video services, online gaming, and e-commerce (1-10 Mbps). The 2000s ushered in fiber optic technology, drastically improving internet speeds, supplemented by mobile broadband networks like 3G and 4G LTE, enabling high-quality streaming on mobile devices (50 Mbps-1 Gbps).

Currently, global fiber optic infrastructure expansion has made gigabit internet common. In some regions, connections exceeding 10 Gbps are possible, and mobile network speeds have approached those of fiber optics since the advent of 5G. This progress paved the way for ultra-HD streaming, cloud gaming, and IoT applications.

Despite these advancements, some applications still suffer from inadequate internet speed. For instance, cloud gaming fails to deliver satisfactory user experiences, and the future of AR/VR industries hinges on extremely low latency and stable internet environments.

### Blockchain’s Unmet Appetite for Scalability

Unlike the internet, blockchain’s evolution depends on decentralization, censorship resistance, culture, and community. Although now defunct, the Terra network exemplified the power of community building under the ideal of decentralized stablecoins. Abstract chains seek to achieve mass adoption through consumer services.

Nevertheless, blockchain’s role as an underlying infrastructure for various applications demands enhanced scalability. Initial solutions like Polygon and BNB Chain improved Ethereum’s limited scalability, enabling fast environments for DeFi. High-performance Layer 1 networks like Solana and Sui, and Layer 2 solutions like Base and Arbitrum, followed, enabling previously challenging services like order book DEXs.

However, current blockchain networks remain significantly slower than centralized internet solutions. Decentralized maintenance introduces performance bottlenecks due to node interactions, and high hardware requirements contradict decentralization principles. For blockchain to realize its potential as the next internet, supporting diverse services (gaming, social, streaming) without hitches, scalability must improve.

### The Allure of Parallel Processing

Numerous blockchain projects propose transaction parallelization as the solution to scalability. Ethereum’s limited scalability stems from its inherently sequential EVM transaction processing. Projects like Solana, Sui, Aptos, Monad, MegaETH, and Polygon have designed VMs or modified the existing EVM to facilitate parallel processing.

Parallel processing hinges on determining if transactions affect the same state. For example, two transactions minting NFTs from the same contract interact with the same state and must be processed sequentially. Parallel processing methods generally fall into two categories:

#### State Access Method

Before processing transactions, State Access methods determine which state each transaction will reference, allowing parallel processing of unrelated transactions. Examples include:

– **Solana:** Utilizes Sealevel for multithreaded parallel processing, handling transactions with mutually exclusive state references concurrently.
– **Sui:** Similar to Solana, Sui’s object-centric data model allows parallel processing if transactions interact with different objects.

#### Optimistic Execution Method

Optimistic Execution processes all transactions in parallel first and sequentially resolves those with conflicts. Examples include:

– **Aptos:** Uses the Block-STM engine for parallel processing, resolving transaction conflicts sequentially afterward.

Parallelization promises improved performance, but its effectiveness diminishes during high network congestion involving numerous related transactions, such as during popular NFT minting events.

## Insights: Somnia’s Focus on Sequential Execution

### Efficacy of Transaction Parallelization

While transaction parallelization offers benefits, it is not without limitations. MegaETH simulations merging Ethereum blocks into batches revealed performance gains, but not a directly proportional increase due to frequent transaction conflicts.

### Somnia’s Core Functions

Somnia aims to maximize blockchain scalability to enable Web 2.0-level on-chain applications through:

– **Multistream Consensus:** Unlike other blockchains, Somnia’s validators each manage their blockchains (data chains), referenced by a consensus chain to determine block order.
– **Sequential Execution:** An EVM compiler converts EVM bytecode into native code, allowing the CPU to directly execute it.
– **IceDB Database:** Optimizes gas fee calculations and cache for rapid operations.
– **Advanced Compression Techniques:** Minimizes redundant data transmission using streaming compression.

Somnia’s focus on efficient single-core processing achieves remarkable transaction throughput, making it a promising candidate for next-level blockchain scalability.

### Maximizing Sequential Execution

#### EVM Compiling

EVM typically uses a stack-based structure, translating bytecode through an interpreter, which is slow. Somnia’s EVM compiler converts bytecode to native code, enabling near-C++ execution speeds. Native code compilation focuses on frequently called smart contracts, managing resources effectively while achieving millions of TPS in benchmarks.

#### Hardware-Level Parallelism

By converting EVM bytecode to native code, CPU’s internal instruction-level parallelism can be utilized. This allows significant performance improvements through simultaneous execution within the CPU core.

## Vision for True On-Chain Applications

While many blockchain projects claim scalability, Somnia aims for Web 2.0 levels, facilitating complete on-chain applications like the metaverse. Future articles will delve deeper into Somnia’s core features and its vision for blockchain-enabled Web 2.0 applications.

## References

Related articles, news, and tweets:
– **Somnia Docs**
– **All About Parallelization by Sui Foundation**
– **Solana Mega Report V2 – Like Apple, but Unlike Apple by Four Pillars**
– **Complete Guide to Sui by Four Pillars**
– **A Monster Combined With Narrative and Technology, Monad by Four Pillars**