Network profile

Solana: high throughput with a different failure mode

Solana buys speed with hardware. That makes it cheap for users and demanding for validators, and it shifts where congestion appears.

Layer 1 · PoS + Proof of HistoryChecked 8 Nov 2026

Solana chooses throughput: it is built to keep one global state moving fast, and it pays for that with hardware requirements most chains never ask of their validators.

Native token
SOL, used for fees, staking and the deposit that keeps account data on chain
Type
Layer 1 with a single global state and no rollup layer beneath it
Consensus
Proof of History as a shared clock, plus proof of stake for leader selection and voting
Fees
A fixed base fee per signature plus an optional priority fee priced in compute units
Settlement
Self-settled; a block becomes «confirmed» and later «rooted» as stake votes accumulate
Ecosystem
Consumer apps, payments, DeFi, DePIN hardware networks, NFTs, trading infrastructure
Explorer
solscan.io
Documentation
solana.com/docs

How it works

Solana's model is unusual in two places: time and state layout.

Proof of History is not a consensus mechanism on its own. It is a verifiable delay: each validator in the leader schedule repeatedly hashes its own output, producing a sequence that proves a certain amount of real time passed between two events. Because the sequence can be checked quickly and only produced slowly, it gives the network a clock that does not need to be agreed on message by message. Blocks then carry a timestamp anyone can verify.

Stake decides who produces blocks. Leaders are scheduled in advance, and validators vote on the blocks they see; votes accumulate weight until a block is rooted. Finality therefore arrives in stages rather than as a single event, and the difference between «confirmed» and «finalised» matters if you are crediting a deposit.

The account model is the other half. Programs hold no state of their own: they are code that reads and writes accounts it does not own. A transaction must declare, up front, every account it will touch, which lets the runtime identify non-overlapping transactions and execute them in parallel. This is the design that makes the throughput possible, and it also makes transactions bigger and the developer model stricter than on an account-based EVM chain.

Storage is paid for with a refundable deposit per account, sized by the data it holds, which keeps state from growing for free and makes «rent» a real concept when you build.

Fees: the mechanism

Two components, and they behave differently:

  • Base fee. A small fixed amount per signature, set by the protocol. It is not an auction, so it does not rise when the network is busy.
  • Priority fee. You attach a price per compute unit and a compute limit; the fee you pay is that price times the units consumed. This is the part that moves with demand.

Because the base fee is fixed, congestion shows up as failed or dropped transactions and as priority fees climbing, not as a rising base price. Priority fees are also applied per account rather than globally, so a popular token or NFT mint creates a local fee spike while unrelated activity stays cheap.

Wallets let you set the priority fee, usually with a preset. Setting it to zero is allowed: you then compete on arrival order alone.

What is actually built there

Solana's centre of gravity is consumer-facing: wallets with embedded onboarding, payment apps, trading interfaces, NFT platforms, and a cluster of hardware networks that use tokens to coordinate physical infrastructure. Low fees and fast confirmation make small, frequent interactions viable, which is why the ecosystem skews towards things people tap rather than things institutions sign.

Trading and market-making infrastructure is deep, partly because the performance profile suits order books, which are awkward on slow chains with expensive computation.

The trade-off is visible in what is scarcer: the shared liquidity and the number of independent node operators are smaller than on Ethereum, and applications that need maximum neutrality or heavy composability across many unrelated protocols have less to work with.

Risks and limits

  • Hardware requirements are high, which pushes validation towards professional operators and data centres. Cheap for users, concentrated for the network.
  • The chain has stopped producing blocks before, and restarts required coordinated action from operators. That is a real property of the design, not a rumour.
  • Congestion appears as failure. When demand spikes, transactions can be dropped rather than simply delayed, which is a worse experience for a payment than a high fee.
  • A single global state means one hot spot affects everything. There is no rollup boundary to absorb a popular application.
  • Wallet drainers and malicious approvals are the dominant user-side loss, as on any account-based chain with token approvals.
  • Client and validator concentration remain open questions.

Official sources