440,000 Transactions per Second With ‘Red Belly’ Blockchain

[Vladdirescu87]

The University of Sydney has recently announced that it has developed a new Blockchain-based system that can perform more than 440,000 transactions per second.

Researchers from the School of Information Technologies at the University of Sydney, who developed the new Blockchain, named this system after one of Australia's most deadly snakes – ‘Red Belly’ Blockchain, and claim that it has a game-changing potential for the financial transfers industry.

Read the full article: https://coinidol.com/440000-transactions-per-second-with-red-belly-blockchain/

[bathrobehero]

I'm curious about their approach.

Very high TX/seconds can be achieved easily but it introduces problems such as rapid blockchain bloat (the size of the blockchain grows rapidly to gigabytes or even terrabytes of data) or too short block times that can cause forks or very high block sizes which require tons of bandwidth excluding regular users to be full nodes.

Even if each transaction would only be 100 bytes each (Bitcoin has about ~3 times bigger transaction sizes on average), each block would use ~42 MB of disk space. That's 42 MB per second. Even one hundredth of that would require a continous bandwidth of 429 KB/s. And it likely wouldn't be perfectly continuous at all so you would need massive speed spikes not to lag behind the network.

And that amount of transaction surely requires a lot of CPU power to actually process.

[Fuserleer]

Known about this before it hit the mainstream news, was waiting for it to show up here.

I was quite excited until I read "consortium of nodes"...then I put the paper down.

Consortiums, Validators, Masternodes....meh....all the same trick.  Reducing the decentralization and attack resistance to increase scalability.  It's also targeted more towards private networks rather than public, so no good for us here.

400k TPS however is quite impressive even for such a system.

[Fuserleer]

I'm curious about their approach.

Very high TX/seconds can be achieved easily but it introduces problems such as rapid blockchain bloat (the size of the blockchain grows rapidly to gigabytes or even terrabytes of data) or too short block times that can cause forks or very high block sizes which require tons of bandwidth excluding regular users to be full nodes.

Even if each transaction would only be 100 bytes each (Bitcoin has about ~3 times bigger transaction sizes on average), each block would use ~42 MB of disk space. That's 42 MB per second. Even one hundredth of that would require a continous bandwidth of 429 KB/s. And it likely wouldn't be perfectly continuous at all so you would need massive speed spikes not to lag behind the network.

And that amount of transaction surely requires a lot of CPU power to actually process.

They talk about account balances being verifiable without requiring the entire ledger...though they don't detail how anywhere.

I would assume that it's doing something like Open Transactions, where each transaction includes the previous balance and the amount.  From that you can determine the current balance, and therefore run heavy pruning.