Synapse Summit — MGEN Node Stress Tests & TPPS Results
We showcased our blockchain with a live stress test at the Synapse Summit. The annual Synapse Summit was held on January 23rd and 24th, 2019 in Tampa, FL which brought together the best ideas and technologies for two days of learning, exploring, and celebrating the possibilities of technology.
The main purpose of the test was to find out the speed at which we can process transactions at whilst increasing the number of nodes. The results of this stress test will help us quantify the effects of transaction processing whilst increasing the number of nodes in the network and stressing the transaction queue. The setup consisted of 6 nodes, we plan to carry out more tests with a higher node count in the future.
Initially we noticed when we sent transactions the node processed them instantly, so our transactions processed per second was limited on the throughput we had sent. So for the purposes of the stress test we made the node wait for us to finish sending 70,000 transactions, then it will process them all together. This helped maximize the load on the consensus and this way we didn’t need 100 computers sending 70K transactions per second to achieve processing 70K transactions per second.
Notice how in the video the transactions processed per second (TPPS) does not go up instantly. This is because we base our TPPS value on the timestamps of the transaction (start processing and end processing timestamps). As the node displaying the metrics syncs more transactions, it obtains a more accurate max TPPS result.
Here are more results of TPPS scaling as number of nodes increase based on 70K transactions:
3 Nodes — 14,796 TPPS https://i.imgur.com/QYaLfII.png
4 Nodes — 20,969 TPPS https://i.imgur.com/szUfcad.png
5 Nodes — 32,704 TPPS https://i.imgur.com/FuJKpo0.png
6 Nodes — 48,082 TPPS https://i.imgur.com/k2ZZX1u.png
In some of the statistics the number of transactions sent is not equal to the number of transactions processed. This is because we do not count the head transaction of a block. For example, the below image shows that Node 220.127.116.11 has 9952 transactions processed and 9963 transactions sent. This means that 11 transactions were used to create 11 blocks.
As we increased the number of nodes the amount of transactions processed per second increased. Node clusters in the MGEN system will be limited to N number of nodes, where N will be the optimum size until we start seeing network overhead from having too many nodes talking to each other. Judging from this test we presume the number N is over 100, which would mean over 100 nodes sharing their computing power in a cluster achieving even higher TPPS results. MGEN strives to create a sustainable and scalable network which improves as it grows. In the future we plan to display stress test results for > 1000 nodes and break the 1M TPPS barrier!
We, at MGEN, believe the transactions per second to be vitally important; however, it is less important than the parameters and algorithm nuances by which it was achieved. In other words, it’s the combination of energy efficiency, security, ease of use, reliability/stability that make the feat of speed matter.
MGEN believes making a single unified system easy for all is the secret to adoptable blockchain. We are building an ecosystem to empower the user through our simple and intuitive account portal. The MGEN community will have access to all pertinent information at their fingertips and allow them control of their accounts in ways yet unheard of in the blockchain space.