A public base-layer blockchain system that prioritizes scalability is called Solana. The project’s objective is to offer a platform that lets programmers build decentralized apps (dApps) without having to account for performance limitations. The Proof-of-History (PoH) timestamp scheme in Solana allows for automatically ordered transactions. To further strengthen network security, it also employs a Proof of Stake (PoS) consensus algorithm. Sub-second settlement timeframes, low transaction costs, and compatibility for all LLVM compatible smart contract languages are additional design objectives.

History

In late 2017, Solana’s creator Anatoly Yakovenko published a draft whitepaper describing a novel distributed systems timekeeping method dubbed Proof of History (PoH). The amount of time needed to agree on the transaction order in blockchains like Bitcoin and Ethereum is one of the scalability constraints. The goal, according to Anatoly, was to automate the blockchains’ transaction sorting process. This would be a crucial component in allowing crypto networks to expand far more than they could at the time. Later, Anatoly collaborated with his old coworker from Qualcomm Greg Fitzgerald to create a single blockchain network in Rust that utilized PoH as its “internal clock.” In February 2018, the two published the project’s first internal testnet (with demo) and official whitepaper. Stephen Akridge, a different ex-QUALCOMM employee, proposed that shifting signature verification to graphics processors might significantly boost transaction performance (i.e., scalability). To start the business that would become Solana Labs, Anatoly enlisted the help of Greg, Stephen, and three other people.

Along with the Qualcomm veterans, former Apple engineers were also members of the founding team. To prevent confusion with the Loom Network, an Ethereum Layer-2 scaling solution, they changed the project’s original name, Loom, to Solana. In Q2 2018, Solana Labs started gathering money to create its own cryptocurrency network.

The team raised somewhat more than $20 million between April 2018 and July 2019 through a number of private token offerings. By the end of July 2019, they announced the sales as a single Series A. The fundraising endeavor was carried out in tandem with Solana’s work on the protocol, which through a number of permissioned testnet iterations until the company unveiled their public incentive testnet, known as Tour de SOL, in Q3 2020. Tour de SOL’s initial phase went online in February 2020, and it is still active today with Solana’s Mainnet Beta version.

Soon after receiving $1.76 million in a public token auction held on CoinList, Solana became live on the Mainnet Beta in March 2020. The beta network for the project supported smart contracts and provided simple transaction capabilities. But no staking incentives were included since Solana was still figuring out the timing of its ongoing issue. As of now, the goal is to transition from the beta stage to a more production-ready version in either late 2020 or early 2021. Currently, Solana Labs continues to make significant contributions to the Solana network, and the Solana Foundation assists in funding continuing development and community-building initiatives

Technology

The eight ideas used in Solana‘s very effective blockchain are as follows: Proof of History (PoH): A timer before consensus Tower BFT: A PBFT Turbine variant that is PoH-optimized: Block propagation protocol Gulfstream: Mempool-less transaction forwarding protocol Sealevel: First parallel smart contract run-time ever Pipelining: Transaction processing unit for validation Cloudbreak: Horizontally scaled accounts database Archivers: Distributed ledger storage PoH: A timer before consensus As nodes inside the network cannot trust the timestamp on messages received from other nodes, determining the time and order in which events happened is the largest issue in dispersed networks. By establishing a network-wide source of time that is cryptographically safe, Solana makes an attempt to resolve this problem using Proof of History (PoH). A high-frequency Verifiable Delay Function (VDF) called Proof of History needs a certain amount of consecutive steps to assess yet yields a singular result that can be openly confirmed. Because they can trust the date and message-ordering they’ve received, nodes can build the next block without first coordinating with the whole network. As a result, consensus overhead is decreased.

Tower BFT: A PBFT variant that is PoH-optimized. To reach consensus on network transactions, Solana uses its own consensus algorithm dubbed Tower BFT, which is a PBFT-like algorithm, to operate on top of Proof of History. A node on the network commits to a period of time when they are barred from voting on an opposing fork each time they vote on a certain fork. As they continue to vote on the same fork, this time during which they are locked out lengthens exponentially until they reach a maximum lockout of 32 votes for the same fork. Nodes in the network should continue voting on the fork they believe the supermajority of the network is supporting because they won’t start receiving inflation rewards until they hit this maximum vote lockout.

Turbine: A block propagation technique is called Turbine. Using a distinct but linked protocol named Turbine, Solana transfers blocks (communicates blocks amongst validators) independent of consensus. Turbine is designed for streaming and makes extensive use of BitTorrent. A block is split up into several random peers as it is transmitted, accompanied with tiny packets containing erasure codes. 

Gulf Stream: A transaction forwarding mechanism without a mempool Mempool management creates a new set of issues in high-performance networks because to the increasing expense of filtering and retaining unconfirmed transactions. Gulf Stream works by moving transaction forwarding and caching to the network’s edge. In the Solana design, clients and validators pass transactions to the anticipated leader in advance since each validator is aware of the order of incoming leaders (block producers). As a result, the unconfirmed transaction pool puts less strain on validators’ memory by enabling them to process transactions in advance, speed up leader switching, and minimize confirmation times. The network gives validators more time to coordinate, which increases the danger of collusion as a result of knowing leaders in advance. Solana’s fast lock time can reduce the possibility of collusion by limiting the time available to plan an attack.

Sealevel: Solana created Sealevel, a highly parallelized transaction processing engine that scales horizontally over GPUs and SSDs. It is a run-time for parallel smart contracts. Single-threaded processors are the norm for other blockchains. In contrast, Solana can enable the processing of simultaneous transactions (along with the verification of signatures) within a single shard. This problem’s solution makes extensive use of the scatter-gather operating system driver approach. The state that transactions will read and write while executing is specified in advance. When reading from and writing to an array of RAID 0 SSDs, Sealevel may identify the non-overlapping transactions occuring in a block and execute them concurrently (referred to as parallel execution). A virtual machine (VM) called Sealevel is used to plan transactions but not carry them out. Instead, utilizing a bytecode known as the Berkeley Packet Filter, Sealevel sends transactions to be processed directly on hardware (BPF).

Pipeline: A TPU for validation and optimization The Solana network’s transaction validation procedure takes considerable use of pipelining, a CPU design improvement. When a stream of input data needs to be processed by a series of stages and separate hardware is in charge of each step, pipeline processing is the best option. The Transaction Processing Unit (TPU), a pipeline mechanism on the Solana network, moves from Data Fetching at the kernel level to Signature Verification at the GPU level, Banking at the CPU level, and Writing at the kernel space level. Before the TPU begins sending blocks to the validators, it has already fetched the following series of packets, checked their signatures, and started crediting tokens. This four-stage pipeline’s GPU parallelization enables the Solana TPU to function at a high performance.

Cloudbreak: Horizontally scaled accounts database, cloudbreak These additional technologies’ methods for scaling computing could lead to memory problems. Because of its small size and slow access rates, memory—which is required to keep track of accounts—can struggle to sustain performance. Cloudbreak was created to be optimized for simultaneous reads and writes distributed across a RAID 0 array of SSDs. Each extra disk increases the amount of storage that on-chain applications have access to, as well as the number of reads and writes that may be executed simultaneously by programs. This complements Solana’s transaction architecture, which permits pre-fetching accounts from disk and setting up the runtime for execution. It also permits nodes on the network to start processing transactions before they are encoded into a block. Everything here strives to decrease network block delays and confirmation latency.

Archivers: Distributed ledger storage for archiving One of the key vectors of centralization is anticipated to be the storing and preserving of data on a high-performance network. Only well-funded businesses will be able to serve as validators and take part in consensus if storage costs are very high. Data storage on Solana is transferred from validators to an archiver network of nodes. Consensus is not engaged by archivists. The state’s history is fragmented into many periods and erasure codes. Small pieces of the state are kept by archivers. The network may occasionally require the Archivers to provide documentation demonstrating that they are properly storing the data. Proofs of Replication (PoRep), which are primarily based on Filecoin, are used by Solana. Archivers are not in use right now but are planned for the future. They will be implemented in response to network data requirements.

Supply Curve Details

Vesting Schedules Each of Solana‘s three pre-launch private sales included a nine-month lockup following the network’s debut. The first trading day for the tokens issued in these transactions is expected to be January 7, 2021. The SOL tokens issued in the project’s public auction sale, which took place in March 2020 and represented 1.6% of the original supply, were completely liquid once the network became live. The project’s public auction sale did not include a lockup schedule. Additionally, a nine-month lockup will be in place following the launch of the network for the founder’s allotment (12.5% of the original supply). These tokens will vest monthly for an additional two years following the lockup period (expected to fully vest by Jan. 2023). Together, the Solana Foundation’s Grant Pool and Community Reserve hold around 38% of the initial SOL supply. Since the introduction of Solana’s mainnet, these allocations have started to vest in modest quantities. For the first nine months, one million Grant Pool tokens vested every three months, and for the next nine months, 35 million Community Reserve tokens vested in five million token increments. In January 2021, the complete allotment for these categories will become available.

Following the introduction of the network, the Solana Foundation’s allotment started vesting in modest amounts each month. At launch, the foundation also made 11,365,067 SOL available for a six-month loan to a market maker. The team decided to burn an equal number of tokens from the foundation’s allotment after informing SOL holders of this information. The “pico-inflation” option was approved by ongoing emissions Solana validators on December 24, 2020. New SOL coins are presently being created at a rate of 0.1% per year, and they are being distributed to validators and stakers in accordance to the stakes they have made. This comparatively low inflation rate is merely transitory; at some point in 2021, it will be replaced by a more significant and long-term inflation idea. The first annual inflation rate in this “final” inflation proposal is set at 8%. The “dis-inflation rate” will, however, see a 15% yearly decline in this inflation rate. The network should reach this level in around 10 years, or in 2031, when Solana’s inflation rate will continue to drop until it achieves an annual rate of 1.5 percent. Unless the governing structure of the network chooses to modify it, Solana’s long-term inflation rate of 1.5 percent will remain the same.

Consensus Details

To cut down on communication costs and latency, Solana’s Proof of Stake (PoS)-based consensus process, known as Tower BFT, uses the network’s Proof of History (PoH) method as a clock before consensus. Voting on each given fork by a validator is limited to a slot, which is a predetermined time period of hashes. One slot typically lasts 400 milliseconds (ms) in the present network configuration. The network has a possible rollback point every 400ms, but with each vote after that, the network would have to wait twice as long to unroll that vote. In other words, secondary votes make it far more difficult to reverse the transactions that were made in a certain slot. As a result, a block with many votes has a better probability of staying in the chain forever. For instance, in the last 12 seconds, each validator cast 32 votes. A delay of 232 slots, or nearly 54 years, will be applied to the vote from 12 seconds ago. In reality, the network won’t ever roll back its decision. The timeout for the most recent vote is two slots, or around 800 milliseconds. Existing blocks are more likely to be confirmed when new blocks are added to the ledger because every slot that old votes are committed to doubles in size. Once two-thirds of the network validators have chosen a certain order of events, Tower BTF gives finality. Transactions cannot be turned back once they have been completed.

Delegated Proof-of-Stake (dPoS), which allows token holders can participate in the block creation process and receive rewards by either staking tokens and becoming their own validators or by delegating their tokens to validators they trust, is the operating model for the Solana mainnet. Any person can join the network as a validator and boost the protocol’s overall security. Although the leader selection procedure (which validator gets to propose the next block) is stake-weighted, there is no minimum staking requirement. The network architecture uses GPU cores to parallelize execution and shorten verification times, and it is built to grow with hardware and bandwidth. Hardware requirements are a little higher than for certain protocols because of the GPU need. An adequate setup should cost no more than $5,000.