“The Casper aspect of Casper, where two conflicting blocks should not be able to be finalized, seems to actually happen running and the network agrees all along what the final chain is. So that part has been totally successful,” Vitalik Buterin, Ethereum’s co-founder, said in a research update during the developers meeting.
The Proof of Stake hybrid Casper is on testnet, and has been running for a few weeks, but not without problems. The nodes apparently keep reconnecting and randomly get a flood of connections, so devs have to perform round-about hacks to keep it going.
That suggests quite a bit more testing will be needed before it can be implemented into the live network. But the main part of Casper, it’s ability to reach network consensus, has apparently been successful, with the faced problems more ancillary. A live implementation, therefore, might hopefully be ready by summer.
“On the Sharding side, quite a lot of work has started,” Buterin said, with sharding being probably the biggest upgrade of any public blockchain.
Currently, blockchains work by every node replicating the exact same work as every other node. That doesn’t scale very well. Parallelization, therefore, will need to be incorporated, which is in effect what sharding does.
So instead of having an identical replication, nodes are bundled into a set of say 100 and validate say 1,000 transactions. The other 100 nodes validate a different set of 1,000 transactions and so on.
Then, through a consensus algorithm, all these nodes are bound together, so that if a node produces an invalid transaction the entire network knows about it and rejects it.
This can allow for as good as an unlimited number of transactions, with the full specification for part one of phase one now at a “good enough” stage.
There are five phases in total, and it is only part one of phase one that is completed, that being the specification. All of that will now have to be fleshed out and coded out, then tested, then implemented.
They do however appear to be moving faster than expected, so with some luck the implementation of the first stage might be even by next year.
In the meantime, the network has been facing some slight congestion, although fees have fallen back below ten cent.
A limit of sorts has been reached at around 1.2 million transactions a day, with miners keeping capacity there because of an increase number of blocks found at the same time, causing one of those blocks to be worth less and thus potentially reducing miners’ revenue.
As developers have focused more on longer term increased capacity through, to some extent, Casper and more fully through Sharding, efficiency improvements were given less priority, until now.
— Péter Szilágyi (@peter_szilagyi) January 19, 2018
That’s an eth developer, with geth being ethereum’s main node software client. The increased efficiency should now make running a node magnitudes easier, which might allow miners to increase capacity unless other bottlenecks are found.
So Ethereum is scaling, with the network managing to keep fees at pennies, while handling the highest level of transactions out of any public blockchain, whether centralized or decentralized.
Clarification: Ethereum researcher Hsiao-Wei Wang said “there may be 4 Phases for long-term and 4 Stages in Phase 1 for short-term. Stage 1 is almost done, and the testing plus some integration works should be finished in one month. (At the meantime, we’re starting Stage 2 in parallel.)”