What are the most impactful specs for a node? And why is the TPS is inconsistent?

Question

Hello,

I've started some TPS tests using Apache Benchmark Tool to run a send command between two different addresses for each node of following devices:

1. VPS, Intel i7 2.4GHz 4 vCore(s), 24 GB RAM, 100GB HD (sender)
2. VPS, Intel i7 2.4GHz 4 vCore(s), 24 GB RAM, 100GB HD (miner)
3. VPS, Intel i7 2.4GHz 2 vCore(s), 12 GB RAM, 50GB HD (miner)
4. VPS, Intel i7 2.4GHz 2 vCore(s), 12 GB RAM, 50GB HD (sender)
5. VPS, Intel i7 2.4GHz 2 vCore(s), 12 GB RAM, 50GB HD (sender)
6. Laptop Intel i7 8th Gen 1.8GHz 4 Core(s), 16 GB RAM, 500 HD (sender)
7. Laptop Intel i7 8th Gen 1.8GHz 4 Core(s), 16 GB RAM, 500 HD (sender)
8. Laptop Intel i7 7th Gen 2.8GHz 4 Core(s), 16 GB RAM, 500 HD (sender)

The tests went out as the following (all of the results are the average of the sender nodes divided by the # of seconds):

1. 1 sec: 3347 TPS
2. 60 sec: 1879 TPS
3. 60 sec: 840 TPS
4. 60 sec: 1095 TPS
5. 300 sec: 625 TPS
6. 300 sec: 313 TPS
7. 300 sec:474 TPS

At first I thought the CPU had to do with the highest TPS for a single device, but these results are also inconsistent per device as well.

• Which specs affect the mining the most?
• Which specs affect the sending the most?
• Why is the TPS results inconsistent?

Comments

I'm curious as for how are you testing your network? Could you share what tools you are using or what's your approach?

---

using Apache Benchmarking Tool to preform a single asset transfer over varied amount of time and throughout different devices as single nodes or in combination

---

oh, i didn't know that tool! i'll definitely check it out. do you make a script with rpc commands that it executes during the test? how did you measure the tps?

Answer 1

Please see this page which has a lot of information on performance optimization:

https://www.multichain.com/developers/performance-optimization/

My guess is your issue is with the number of UTXOs in wallets.