Solana Trader API: Stake-weighted Quality of Service (SWQoS) Performance Benchmark
By: Eyal Markovich, Co-founder & COO
Recently, TraderAPI introduced the “useStakedRPCs” option, integrating SWQoS connections to route transactions more efficiently. This partnership with several validators aims to enhance transaction success via SWQoS, though its effectiveness is still under scrutiny. In this post, we provide a snapshot of our ongoing evaluation and benchmark results.
Current Status of SWQoS Implementation
- Success rate setting up connection: So far, 40% of validators we partnered with have successfully implemented SWQoS. This indicates that enabling SWQoS is complex and has varying success rates.
- Challenges: Many validators have to disable the Jito relayer to allow SWQoS to function, though this does not guarantee success. The variability suggests environmental factors impacting SWQoS performance. Our team continues to investigate the conditions under which swQOS operates effectively.
Benchmarking SWQoS
To understand SWQoS’s impact, we conducted tests with and without the “useStakedRPCs” flag under identical conditions. Both tests were submitted as a Trader API Enterprise User, and used the following settings:
- FrontRunningProtection = false
- FastBestEffort = false
- Tip = 0.0005 SOL
- Priority Fee = 0.0005 SOL
The only difference between the two tests was the value of the “useStakedRPCs” parameter.
Setting “useStakedRPCs” to ‘true’ routes a transaction in multiple SWQoS connections as well as uses more nodes to propagate to the leader. Because of this, we can verify that the performance improvements in the test are attributed solely to SWQoS.
Test 1: Control (without SWQoS)
- Transactions: 100
- UseStakedRPCs=false
- Result: 100% success, with a 90th percentile completion time of 6.7 seconds.
Test 2: with SWQoS
- Transactions: 100
- UseStakedRPCs = true
- Result: 100% success, with a 90th percentile completion time of 3.6 seconds, demonstrating a significant improvement.
Cost of Experiments: 6 SOL.
The tests indicate that when “useStakedRPCs” is used it substantially reduces transaction landing latency, enhancing throughput and efficiency.
We have repeated the tests several times, with consistent results. There is a noticeable improvement between using “useStakedRPCs” and not using it. We also observed improvements in transaction landing latency and throughput when using an Elite plan (as opposed to an Enterprise plan). . The actual times to land the transaction vary with changes in tip, priority fee, and network congestion.
Conclusion
While the potential of SWQoS in improving transaction execution is evident, its current inconsistent performance across different environments calls for further optimization and testing.
For more detailed insights and continuous updates on our benchmarks, please refer to our Solana Trader API documentation and future blog posts.
