In 2022, as the popularity of SBF and FTX platforms soared, Solana also became a hot public chain in the crypto industry, but then the collapse of FTX almost brought down the entire Solana ecosystem.

The price of SOL plummeted from $236 to $13 in a few weeks. Investment institutions advised startups not to choose Solana and instead build on the Ethereum Virtual Machine (EVM). Subsequently, some well-known projects migrated from Solana to other chains. However, a year later, as shown in the figure below, Solana rebounded and surpassed its peers.

People like to hear stories of comebacks. Today, this article explores what choices Solana made to successfully recover from the bear market trough in 2023, and where the design differences are that made it a head public chain.

Solana's founder Anatoly and team members have rich backgrounds in the traditional mobile communications industry. They worked at Qualcomm as programmers for more than a decade and witnessed the impact of Moore's Law (hardware capacity doubles every two years). Solana is built differently than Bitcoin and Ethereum, and it does not limit the requirements of node hardware.

Bitcoin and Ethereum are relatively mature networks with more client diversity. But why is client diversity so important? Think of it this way, a decentralized network, you want all functions to be relatively decentralized, if more than 66% of the network uses a single node client, and that node issues incorrect updates or chooses to sync blocks in the wrong order, it will affect the functionality of the blockchain. There may be consensus issues about which block is approved first, and both Ethereum and Bitcoin have actively optimized client diversity in the past.

Solana experienced three major network failures and several performance degradations in 2022, and one failure in 2023. These outages were mainly caused by consensus issues. Although low transaction fees are good for users, it also makes it easier to attack by sending a large number of transactions or conducting denial of service (DDoS) attacks.

When a block is proposed, validators receive the information packet (in the block), independently verify its correctness, and confirm the correctness with each other to reach consensus. However, when validators lag in processing the information packet, the consensus information is lost.

Firedancer has created a messaging framework that bypasses certain hubs and reduces latency in the network. Since Firedancer is built from scratch by a different team, it may not carry the same bugs as the Solana Labs client. Therefore, the same bug will not affect these clients at the same time. Ideally, validators will run a primary and a secondary client, with the secondary client as a backup.

A chain with a strong DeFi ecosystem needs to guarantee 100% uptime, so Solana needs a more robust client infrastructure. The main reasons for the Solana network stalling are the lack of congestion control and network processing delays. Several network upgrades have improved the validator's anomaly to the transaction flood and achieved better congestion control.

Solana admits that client diversity is a work in progress. As with Ethereum and Bitcoin in the past, these things take time. One sign of improvement is the percentage of assets running through the Jito-Solana client. While the Jito Solana client does not help in achieving redundancy, it suggests that validators will run different clients when available.

As more clients like Firedancer and Sig come online, we should see less reliance on the Solana Labs client in the future. The best percentage for individual clients is around 33%. So there is still work to be done.

A healthy fee market is a key factor for a thriving blockchain, as shown by chains like Bitcoin and Ethereum. In 2024, Bitcoin’s block reward will halve from 6.25 BTC to 3.125 BTC per block. If we assume that Bitcoin producers require the same incentives, then for existing incentive levels to be maintained, the price must double, or fee revenue must make up for the loss of the halved reward. Thanks to inscriptions, increased fees bring hope to block producers and Bitcoin security budgets.

Through EIP1559, Ethereum changed its monetary policy, ensuring that ETH's inflation remains under control by adding a destruction mechanism. Monetary systems and dynamic fees play an important role in stabilizing the chain and aligning stakeholder incentives, and other chains also hope to achieve the same status.

Solana has no priority fees in its initial stage, and the fee for each transaction is fixed at 5000 Lamports (in the Solana blockchain, Lamports is the smallest unit, similar to wei in Ethereum or satoshis in Bitcoin). Solflare is the first wallet to implement priority fees on Solana in January 2023. Fees are crucial for the following reasons:

1. Protection against spam attacks

2. Validator rewards

3. Improvements to the economic stability of the protocol. As fees increase, inflation can be reduced.

Like Ethereum's EIP1559, Solana destroys 50% of fees, and the remaining 50% belongs to the validators. This standard was set in 2021 and has not changed to this day.

On Ethereum, transactions wait in the mempool before entering a block, and validators select the transactions that pay the highest fees for block packaging. The global mempool is created by different validators broadcasting their respective mempools to each other. This is where the maximum extractable value (MEV) is generated.

Because the mempool is visible to validators and MEV searchers, searchers can identify transactions that can be pre- and post-traded for profit. Searchers are usually robots that look for MEV opportunities. For example, if someone buys token A worth one million dollars, a searcher can buy A before that transaction is completed and sell it immediately.

Unlike Ethereum, Solana is multi-threaded and can execute transactions in parallel. When signed transactions arrive at the leader, the leader verifies them and randomly assigns them to threads. Only when assigned to different threads local to the leader are they sorted by priority fee (i.e., transactions with the highest fees go first).

Solana originally had no priority fees. But now, wallets like Solflare allow users to pay priority fees. Priority fees give rise to Solana’s local or segregated fee market. Unlike Ethereum, Solana transactions must specify where they wish to read and write a portion of the state.

Solana’s validators know the state involved in the transaction before the computation, while Ethereum validators know this only after they start the computation. Solana transactions need to specify specific information that helps Solana determine which part of the state is becoming a hotspot. The total number of compute units (CUs) used by any hotspot is limited to 25% (of one of Solana’s four cores for multi-threaded execution). This is done to prevent an account from being updated too many times in a single block.

A hotspot is a specific smart contract or account that suddenly sees a lot of traffic. On the EVM network, a large demand for a single application (such as Crypto Kitties) can cause transaction fees to rise for the entire network. On Solana, the number of CUs that individual smart contracts/applications (such as Tensor or Jupiter) can use per block is limited to 25%.

That is, transactions using any particular contract cannot take up more than 25% of a block, or 12 million CUs. All transactions above this limit must wait for the next block. So if a standalone application sees a dramatic increase in usage, the entire network won't start paying more fees. Only transactions interacting with that application will see an increase in fees. This is what a localized fee market looks like.

What happens if there are 4 or more hotspots? In this case, Solana looks like Ethereum. Gas wars could occur between competing hotspots, and the transactions with the highest fees will get in. Local fee markets seem like a beautiful solution to the general problem of fee surges.

How does it work in practice? There are still some problems with Solana's fee market design:

First, currently transactions incur the same base fee, whether it's a token transfer, an exchange, or a flash loan. This is obviously not reasonable. Transactions should incur fees based on the computational resources (CU) consumed, although this is already under consideration. CU represents block space, so paying higher fees should be able to get more space.

Second, since there is no mempool, validators will only schedule transactions based on fees after assigning them to different threads, so transactions with higher fees will not always succeed. This can lead to the next problem.

Third, Solana does not have a mempool like Ethereum, so higher priority fees do not guarantee that the transaction will be included in the block. So, the best way for a seeker (someone looking for MEV) to extract MEV is to bombard the network with multiple transactions and hope that the validator chooses one of them. On Solana, this is relatively easy to do due to the low transaction costs.

In 2022, as the popularity of SBF and FTX platforms soared, Solana also became a hot public chain in the crypto industry, but then the collapse of FTX almost brought down the entire Solana ecosystem.

The price of SOL plummeted from $236 to $13 in a few weeks. Investment institutions advised startups not to choose Solana and instead build on the Ethereum Virtual Machine (EVM). Subsequently, some well-known projects migrated from Solana to other chains. However, a year later, as shown in the figure below, Solana rebounded and surpassed its peers.

People like to hear stories of comebacks. Today, this article explores what choices Solana made to successfully recover from the bear market trough in 2023, and where the design differences are that made it a head public chain.

Solana's founder Anatoly and team members have rich backgrounds in the traditional mobile communications industry. They worked at Qualcomm as programmers for more than a decade and witnessed the impact of Moore's Law (hardware capacity doubles every two years). Solana is built differently than Bitcoin and Ethereum, and it does not limit the requirements of node hardware.

Bitcoin and Ethereum are relatively mature networks with more client diversity. But why is client diversity so important? Think of it this way, a decentralized network, you want all functions to be relatively decentralized, if more than 66% of the network uses a single node client, and that node issues incorrect updates or chooses to sync blocks in the wrong order, it will affect the functionality of the blockchain. There may be consensus issues about which block is approved first, and both Ethereum and Bitcoin have actively optimized client diversity in the past.

Solana experienced three major network failures and several performance degradations in 2022, and one failure in 2023. These outages were mainly caused by consensus issues. Although low transaction fees are good for users, it also makes it easier to attack by sending a large number of transactions or conducting denial of service (DDoS) attacks.

When a block is proposed, validators receive the information packet (in the block), independently verify its correctness, and confirm the correctness with each other to reach consensus. However, when validators lag in processing the information packet, the consensus information is lost.

Firedancer has created a messaging framework that bypasses certain hubs and reduces latency in the network. Since Firedancer is built from scratch by a different team, it may not carry the same bugs as the Solana Labs client. Therefore, the same bug will not affect these clients at the same time. Ideally, validators will run a primary and a secondary client, with the secondary client as a backup.

A chain with a strong DeFi ecosystem needs to guarantee 100% uptime, so Solana needs a more robust client infrastructure. The main reasons for the Solana network stalling are the lack of congestion control and network processing delays. Several network upgrades have improved the validator's anomaly to the transaction flood and achieved better congestion control.

Solana admits that client diversity is a work in progress. As with Ethereum and Bitcoin in the past, these things take time. One sign of improvement is the percentage of assets running through the Jito-Solana client. While the Jito Solana client does not help in achieving redundancy, it suggests that validators will run different clients when available.

As more clients like Firedancer and Sig come online, we should see less reliance on the Solana Labs client in the future. The best percentage for individual clients is around 33%. So there is still work to be done.

A healthy fee market is a key factor for a thriving blockchain, as shown by chains like Bitcoin and Ethereum. In 2024, Bitcoin’s block reward will halve from 6.25 BTC to 3.125 BTC per block. If we assume that Bitcoin producers require the same incentives, then for existing incentive levels to be maintained, the price must double, or fee revenue must make up for the loss of the halved reward. Thanks to inscriptions, increased fees bring hope to block producers and Bitcoin security budgets.

Through EIP1559, Ethereum changed its monetary policy, ensuring that ETH's inflation remains under control by adding a destruction mechanism. Monetary systems and dynamic fees play an important role in stabilizing the chain and aligning stakeholder incentives, and other chains also hope to achieve the same status.

Solana has no priority fees in its initial stage, and the fee for each transaction is fixed at 5000 Lamports (in the Solana blockchain, Lamports is the smallest unit, similar to wei in Ethereum or satoshis in Bitcoin). Solflare is the first wallet to implement priority fees on Solana in January 2023. Fees are crucial for the following reasons:

1. Protection against spam attacks

2. Validator rewards

3. Improvements to the economic stability of the protocol. As fees increase, inflation can be reduced.

Like Ethereum's EIP1559, Solana destroys 50% of fees, and the remaining 50% belongs to the validators. This standard was set in 2021 and has not changed to this day.

On Ethereum, transactions wait in the mempool before entering a block, and validators select the transactions that pay the highest fees for block packaging. The global mempool is created by different validators broadcasting their respective mempools to each other. This is where the maximum extractable value (MEV) is generated.

Because the mempool is visible to validators and MEV searchers, searchers can identify transactions that can be pre- and post-traded for profit. Searchers are usually robots that look for MEV opportunities. For example, if someone buys token A worth one million dollars, a searcher can buy A before that transaction is completed and sell it immediately.

Unlike Ethereum, Solana is multi-threaded and can execute transactions in parallel. When signed transactions arrive at the leader, the leader verifies them and randomly assigns them to threads. Only when assigned to different threads local to the leader are they sorted by priority fee (i.e., transactions with the highest fees go first).

Solana originally had no priority fees. But now, wallets like Solflare allow users to pay priority fees. Priority fees give rise to Solana’s local or segregated fee market. Unlike Ethereum, Solana transactions must specify where they wish to read and write a portion of the state.

Solana’s validators know the state involved in the transaction before the computation, while Ethereum validators know this only after they start the computation. Solana transactions need to specify specific information that helps Solana determine which part of the state is becoming a hotspot. The total number of compute units (CUs) used by any hotspot is limited to 25% (of one of Solana’s four cores for multi-threaded execution). This is done to prevent an account from being updated too many times in a single block.

A hotspot is a specific smart contract or account that suddenly sees a lot of traffic. On the EVM network, a large demand for a single application (such as Crypto Kitties) can cause transaction fees to rise for the entire network. On Solana, the number of CUs that individual smart contracts/applications (such as Tensor or Jupiter) can use per block is limited to 25%.

That is, transactions using any particular contract cannot take up more than 25% of a block, or 12 million CUs. All transactions above this limit must wait for the next block. So if a standalone application sees a dramatic increase in usage, the entire network won't start paying more fees. Only transactions interacting with that application will see an increase in fees. This is what a localized fee market looks like.

What happens if there are 4 or more hotspots? In this case, Solana looks like Ethereum. Gas wars could occur between competing hotspots, and the transactions with the highest fees will get in. Local fee markets seem like a beautiful solution to the general problem of fee surges.

How does it work in practice? There are still some problems with Solana's fee market design:

First, currently transactions incur the same base fee, whether it's a token transfer, an exchange, or a flash loan. This is obviously not reasonable. Transactions should incur fees based on the computational resources (CU) consumed, although this is already under consideration. CU represents block space, so paying higher fees should be able to get more space.

Second, since there is no mempool, validators will only schedule transactions based on fees after assigning them to different threads, so transactions with higher fees will not always succeed. This can lead to the next problem.

Third, Solana does not have a mempool like Ethereum, so higher priority fees do not guarantee that the transaction will be included in the block. So, the best way for a seeker (someone looking for MEV) to extract MEV is to bombard the network with multiple transactions and hope that the validator chooses one of them. On Solana, this is relatively easy to do due to the low transaction costs.