The sudden popularity Cryptococcal demonstrated the limited bandwidth of the Ethereum blockchain, completely blocking the operation of the network with more than 700,000 transactions per day.
The demand for transaction Ethereum is growing steadily for several months, but it seems that only cats have managed to create a critical overload, resulting in mempool increased to 20 000 transactions, and the fee for a simple transfer of air has reached 11 cents.
The situation is complicated by the fact that miners (in particular, Ethermine)
refuse to increase the gas floating-limit (similar to block size in Bitcoin) because the percentage anklav («abandoned» blocks) is too large. Nick Johnson (Nick Johnson) from the Fund Ethereum explains:
The bigger the block, the more anklav; code optimization that increases the speed of propagation of blocks through the network and the speed of processing, will greatly reduce the appearance of anklav at all levels.
Work to improve the processing speed of the units had already been carried out, especially in the midst of a DoS attack on the network last fall, but then the problem was solved hardforum, and the need for modifications depart on the second plan.
I believe such work will resume now: I see at least a couple of things that can be done very quickly.
Recently, the Fund was more involved in the global scaling problem, solving the problem of PoS and sharding, scoring seemingly insignificant problems of miners. However, this seemingly harmless game of cat requires urgent action. Johnson cites some of them:
«The recent proposal EIP on the compression of data in the transaction is able to significantly reduce their volume. This proposal has already been implemented in the client Geth, but the second client Parity, it doesn’t support yet, so while the data exchange between the Geth and the Parity is with the uncompressed data. Besides, Geth have already implemented the entry of the prefix tree with extended drain intervals, instead of writing them in each block. This helps to instantly ignore obsolete data.»
On the way to the simplest solution — increase hashlimit — it is necessary to eliminate several bottlenecks. The first of them is data storage, and this task is already solved. Then, it is necessary to reduce the transmission time between new blocks miners using data compression, according to Johnson, the reason for team Parity.
In addition, very soon the Ethereum implements its own version of FIBRE, xThin or Graphene: it will significantly speed up the communications between the miners, which dramatically reduces the percentage anklav.
Commenting on the refusal Ethermine from increased haslimit, Johnson said, «I think they are right in the fact that the increase in haslimit right now will not help. At the same time, minor modifications can change this situation».
The timing of the decision problem even in the short term to define, especially in regard to the effective distribution of blocks. Johnson made the following assumptions:
«It depends on the teams of both clients. Need to release new versions of the clients and see what effect these changes will have on the appearance anklav; I think that a couple of weeks at least.
I think Cryptococci optimize your contract faster, or the demand for them will fall. Of course, this does not reduce the importance of urgent modifications to customers.»
The seals were only «the last straw», so it is unlikely that the demand for transaction will fall much in the near future.
Acne Buterin wrote on Twitter that the 20 cents per transaction – an unacceptable cost, and most likely it is only a temporary phenomenon, until the modifications, which will be a temporary solution until the implementation of Casper and sharding.
As for the realistic forecast for the next few weeks, then network congestion will continue, and the Commission will raise further, while a full-fledged update customers who can convince the miners to raise haslimit will appear at least at the end of the month, and most likely, early next year.