Why Solana TPS and Congestion Matter for Traders
If you trade on Solana DEXes, you’ve felt it: sometimes swaps are instant and cheap, other times you spam "Approve" and still miss the move. The difference usually comes down to throughput (TPS) and how the network handles congestion.
This article focuses on what actually happens on Solana today (as of 2025–2026), using real data and documented behavior, not marketing numbers. The goal is to help you:
- Read TPS metrics correctly
- Understand why transactions fail during congestion
- Adjust fees and settings so your trades land when it matters
TPS on Solana: Capacity vs. Real Usage
The headline numbers
Solana’s own materials still describe the network as having 400 ms block times and theoretical capacity around 65,000 TPS, with typical transaction fees under $0.001 per transaction.【0search3】 These are design targets under ideal conditions, not what you usually see in live trading.
In practice:
- Independent on‑chain analytics and community reports show average real TPS (including vote transactions) in the low thousands, with non‑vote TPS typically in the high hundreds to low thousands, depending on activity.【0reddit15】
- A 2025 community report summarized that Solana averaged over ~1,100 TPS of actual on‑chain activity per day in 2025, up from ~819 TPS in 2024.【0reddit15】 This aligns with other analytics that show Solana operating at payment‑network scale, not just as a lab demo.【0search11】
The key takeaway: Solana really is high‑throughput compared to most L1s, but the famous 65k TPS is a ceiling, not the day‑to‑day reality.
Vote vs. non‑vote transactions
When you look at TPS charts (Solscan, Helius dashboards, etc.), you’ll often see very high numbers. A big chunk of that is vote transactions:
- Validators constantly send vote transactions to agree on blocks and maintain consensus.
- These are necessary for the network but don’t represent user activity.
For traders, the number that matters is non‑vote TPS:
- Swaps
- Transfers
- Liquidations
- NFT mints, etc.
Many community analyses point out that headline TPS is heavily inflated by votes, while non‑vote TPS is much lower but still high relative to other chains.【0reddit22】
What Congestion on Solana Actually Looks Like
Solana has gone through several well‑documented congestion episodes, especially during periods of intense memecoin and DeFi activity.
April 2024 congestion: a clear case study
In March–April 2024, Solana experienced severe congestion under heavy retail and bot load:
- Non‑vote transaction failure rates spiked above 75% during peak demand, according to technical post‑mortems.【0search6】
- A ChainScore analysis of Solana’s fee model noted that during the March 2024 congestion crisis, TPS remained above 2,000, but failed transactions surged, and users who paid higher priority fees saw >95% success rates.【0search10】
- Anza (a Solana core dev shop) identified issues in the QUIC networking implementation of the Agave validator client as a key bottleneck and published plans to address it.【0search5】【0reddit24】
From a trader’s perspective, this translated into:
- Wallets showing "Transaction failed" or "Blockhash not found"
- DEX swaps needing multiple retries
- Bots and MEV searchers aggressively bidding up priority fees
The important nuance: the chain didn’t stop. Blocks kept producing and TPS stayed high, but capacity was fully saturated, so only well‑priced, well‑formed transactions got through.
Why Congestion Happens on Solana
Solana’s architecture is designed for high throughput, but that also means when demand spikes, you see very sharp contention for blockspace.
Key factors behind congestion:
1. Networking bottlenecks (QUIC)
Solana moved from a custom UDP‑based protocol to QUIC (a modern transport protocol over UDP). QUIC is more robust than raw UDP, but:
- The initial QUIC implementation in the Agave validator client became a bottleneck under extreme load.
- Core contributors (Anza, Jito, Firedancer teams, etc.) have publicly described this as a known bottleneck that needed engineering work to handle the traffic Solana was seeing in 2024.【0search5】【0reddit24】
For traders, this shows up as packet drops and delayed forwarding of transactions to the current leader, which increases failure odds when blocks are full.
2. Stateless spam and bots
Because Solana fees are low by design, spam and aggressive botting are economically viable:
- Arbitrage and sniping bots may fire many transactions per block to win a race.
- During memecoin seasons, mint and swap bots flood the network with high‑frequency orders.
Solana’s fee model is designed to handle this via local fee markets and priority fees (more below), but when demand outruns current optimizations, users paying only base fees suffer.
3. Hot accounts and local contention
Solana parallelizes execution across accounts. If many transactions touch the same accounts (e.g., a single popular Raydium pool or a mint account):
- Those transactions must be serialized, even if global TPS is high.
- This creates local congestion around specific markets or tokens.
This is why you might see:
- Your USDC transfer go through instantly
- But your swap on a hyped memecoin pool fails repeatedly
Even though both live on the same chain, they contend for different resources.
How Solana’s Fee Model Handles Congestion
Solana’s fee system is intentionally different from Ethereum’s global gas auctions. The core ideas:
Base fee + local fee markets
Solana charges a very low base fee per transaction (on the order of fractions of a cent).【0search3】 On top of that, it implements local fee markets:
- Fees are per‑account and per‑resource, not a single global gas price.
- If a particular account (e.g., a DEX pool) is heavily used, fees for touching that account rise, while unrelated activity stays cheap.
A recent technical analysis described this as an "anti‑fragile" fee model: during congestion, the network prefers to drop low‑fee transactions touching hot accounts while keeping overall liveness and throughput high.【0search10】
Priority fees and Jito tips
On top of the base and local fees, users can add priority fees:
- Priority fees are optional extra lamports that signal to validators that your transaction should be included sooner.
- A 2023–2024 update directed all optional priority fees to validators, aligning incentives for them to include higher‑fee transactions.【0search4】
The Jito ecosystem adds another layer:
- Jito provides a MEV‑aware validator client and a TipRouter system that routes user tips (priority fees) to validators.【0search2】
- Jito’s research shows that priority fee usage rises during volatility, and that a growing share of marginal fees flows through Jito when execution quality matters most.【0search9】
For traders, the implication is clear: during congestion, you must pay competitive priority fees (either directly or via Jito‑integrated tools) if you want your transactions to land reliably.
Reading TPS and Congestion as a Trader
1. Don’t over‑interpret raw TPS
When you see Solana TPS charts:
- Remember that headline TPS includes votes.
- Focus on non‑vote TPS and failure rates when available.
- Tools like Helius dashboards, Solscan, and third‑party analytics often break this out.
High TPS with low failure rate is good: the chain is busy but healthy.
High TPS with high failure rate means: you’re in a fee war.
2. Watch failure rates and ping/latency
During the April 2024 congestion, users reported:
- 20–40s average ping times
- 30–50% packet loss
- 50–80% failed transactions in some periods【0reddit14】
You don’t need exact numbers in your wallet, but you should:
- Assume that multiple retries will be needed when the network is under visible stress
- Adjust slippage and priority fees upward during those windows
3. Understand hot spots
A single hyped token can:
- Saturate a specific DEX pool
- Cause local congestion even if global metrics look fine
If your transfers and blue‑chip swaps are fine but your meme pool fails constantly, you’re likely hitting a hot account bottleneck, not chain‑wide failure.
Practical Tips to Trade Through Congestion
Here are concrete, chain‑specific tactics you can use.
1. Always set a reasonable priority fee
On Solana, a tiny increase in priority fee can dramatically improve execution odds during congestion:
- Many wallets and DEX UIs (e.g., Jupiter) expose a "priority" or "fast" setting that automatically increases fees.
- When volatility is high, avoid "min" or "economy" fee presets; choose at least "medium" or "high" priority.
Because base fees are so low, even a 10–50x increase in lamport fee is still cheap in dollar terms but can move you ahead of spam and underpriced bots.
2. Use fewer, better‑constructed transactions
When the network is congested:
- Avoid spamming retries with the same low‑fee transaction.
- Instead, bump the priority fee and send a single, well‑priced transaction with:
- Realistic slippage for the current volatility
- Correct compute budget (if your tool exposes it)
Some advanced tools let you:
- Increase compute units (via
ComputeBudgetinstructions) and pay extra fees for more complex swaps. - Combine multiple steps (e.g., route through Jupiter) in a single transaction to reduce coordination risk.
3. Time your trades around extremes
If you’re not scalping seconds‑level moves, you can often:
- Avoid peak congestion windows (e.g., right at a major airdrop claim opening or hyped mint start)
- Enter or exit positions slightly before or after those peaks when fees and failure rates normalize
This is especially relevant for:
- Retail traders with small size
- Strategies that don’t require millisecond‑level execution
4. Monitor network health alongside price
Before executing large trades, check:
- Solana status pages and dashboards (official status page, Solscan, Helius blog posts about congestion)【0reddit20】
- DEX‑side metrics: some frontends surface recent failure rates or recommended priority fees.
If you see signs of elevated failure rates or delayed block times, treat it like high volatility in infrastructure and size/fee accordingly.
Looking Ahead: Firedancer and Future Throughput
Solana’s roadmap includes major client diversity and performance upgrades:
- Firedancer, an independent validator client by Jump Crypto, is designed to massively increase throughput and reduce reliance on a single client implementation.【0search11】
- Networking and QUIC improvements from teams like Anza aim to remove current bottlenecks that caused the 2024 congestion episodes.【0search5】【0reddit24】
There have also been stress tests and tooling benchmarks showing Solana infrastructure handling well over 100k TPS in controlled scenarios, and even higher transaction streaming rates in specialized tools.【0reddit13】【0reddit17】 These are not everyday conditions, but they demonstrate headroom for future growth.
For traders, the main implication is:
- Expect continued improvements in real non‑vote TPS and reliability over time.
- But also expect that popular markets will always be fee‑competitive—high throughput doesn’t eliminate fee wars; it just raises the ceiling.
Key Takeaways for Solana Traders
- TPS headlines are capacity, not reality. Real non‑vote TPS has been in the hundreds to low thousands, which is still high by L1 standards.【0reddit15】【0search11】
- Congestion doesn’t always mean outages. In recent crises, Solana kept producing blocks at >2,000 TPS, but low‑fee transactions failed at very high rates.【0search10】【0search6】
- Local fee markets and priority fees decide who gets in. During congestion, only transactions with competitive priority fees and correct construction reliably land.【0search10】【0search4】
- Hot accounts matter. A single memecoin pool can be congested even when the rest of the chain feels fine.
- You can trade through congestion by:
- Using medium/high priority fee presets
- Avoiding spammy retries and instead bumping fees
- Timing entries around the worst spikes
- Watching network health metrics, not just price
Understanding how Solana’s TPS and congestion actually work—beyond marketing numbers—gives you a real edge. You’ll waste less time on failed swaps, pay fewer hidden costs in missed fills, and position yourself better for the next wave of activity on the network.