[ad_1]
Open-source Solana implements a high-performance, permissionless blockchain. The Geneva-based Solana Foundation manages the open source project.
On a gigabit network, a centralized database may execute 710,000 transactions per second assuming they average 176 bytes. Optimistic Concurrency Control allows a centralized database to replicate and maintain high availability without affecting transaction throughput.
Solana is showing that blockchain in an adversarial network has the same theoretical restrictions. Key ingredient? Sharing time when nodes are unreliable. When nodes may depend on time, ~40 years of distributed systems study applies to blockchain.
Surprisingly, it can be accomplished utilizing a Bitcoin technique from the start. Bitcoin’s nLocktime feature postsdates transactions using block height instead of a timestamp. If your Bitcoin client doesn’t depend on the network, utilize block height instead of a timestamp. Cryptography calls block height a Verifiable Delay Function. A cryptographically safe technique to mark time. As a more granular verifiable delay function, a SHA 256 hash chain checkspoints the ledger and coordinates consensus in Solana. We now have Optimistic Concurrency Control and are nearing the theoretical limit of 710,000 transactions per second.
Anatoly Yakovenko released a whitepaper on Proof of History in November 2017 to preserve time between untrustworthy systems. Anatoly’s experience creating distributed systems at Qualcomm, Mesosphere, and Dropbox taught him that a reliable clock simplifies network synchronization. Simple synchronization can provide a speedy network limited only by bandwidth.
Anatoly observed as blockchain systems without clocks, like Bitcoin and Ethereum, failed to expand above 15 tps globally whereas Visa needed 65,000 tps. Their failure to become a worldwide payment system or supercomputer was obvious without a clock. When Anatoly solved the difficulty of getting untrusting computers to agree on time, he understood he could apply 40 years of distributed systems research to blockchain. The cluster would be 10,000 times quicker out of the gate!
[ad_2]
Source link