Race to Million TPS: EOS, Ethereum and ThunderChain
Since BlockOne announced the planned launch of EOS on June 2 and finally went through on June 15, EOS with million TPS capability has been a focus of the whole industry.
Vitalik Buterin, co-founder of Ethereum, announced during the 2018 Ethereum Technology and Application Conference in Beijing on June 3 that sharding, as the scalability solution for the Ethereum blockchain network, will be coming soon and allow the network to process millions of transactions per second in future.
Even before that, Chinese Internet veteran Xunlei, had announced in April the launch of ThunderChain, an open blockchain platform supporting millions of transactions per second. Since then, ThunderChain has received attention from the government and the general public.
Although a number of other blockchain programs claims to handle millios of transaction per second, ThunderChain, EOS and Ethereum are the most likely contenders for the forerunners in the Blockchain 3.0 sphere, whether in terms of technological breakthrough or brand reputation.
Blockob, a leading blockchain publication and research institution in China, offers an in-depth analysis of the three millions TPS blockchain platform and explores enormous implications on the future blockchain applications.
ThunderChain gets a head start
Limited TPS is the core reason holding back phenomenal blockchain platforms. The bitcoin network currently processes 5-7 transactions per second and it takes half an hour to confirm a transaction across the network. Theoretically, an Ethereum transaction takes 15 seconds and confirmation is about 3 minutes. These rates are fast enough for the handling of cryptocurrencies but hardly possible to support other application scenarios.
That’s why the market was excited about EOS and its scalability as described in its White Paper 2.0 in last December. EOS had raised up to US$ 4.2 billion before the mainnet launch on June 15.
Ethereum also identifies low performance as a bottleneck for blockchain platforms. Since the beginning of this year, Vitalik has repeatedly stressed the use of sharding in tackling Ethereum’s performance issues and eventually reaching millions of transactions per second.
On April 20, a blockchain platform supporting millions TPS was released by Xunlei. This incredible news was followed by Xunlei’s announcement in May to host the Xunlei Blockchain Application Global Challenge, allowing more developers to get hands-on experience of this millions TPS-enabled blockchain platform.
As shown on the timetable, the mainnet launch of EOS took place on June 15. The current single chain proved to reach 3000+ TPS. According to Tan Zhiyong, CTO and co-founder of OracleChain, the optimized multi-core, non-segmented version to be implemented by the end of this year is likely to achieve millions TPS based on a 128-core machine.
For Ethereum, the goal of millions TPS will take longer. “We hope to get ready in one or two months”, Vitalik said on June 3, adding that his team was making that happen, but the deployment would take a year, so it’s too early to tell. Ethereum is expected to reach millions TPS as soon as next July.
ThunderChain and Ethereum are based on similar architect design. Both Ethereum’s sharding technology and ThunderChain’s homogeneous multi-chain framework allow multi-chain scaling to achieve millions TPS. So, what makes ThunderChain the lead?
“We have sold like over 1 million blockchain devices, which means the owners of such devices – over 1 million households – are part of our ecosystem”, Xunlei CEO Chen Lei said, adding that the size of user base means a very high throughput. As a result, millions TPS-enabled ThunderChain is developed upon users’ demand.
Similarity and difference in architecture
Blockchain is known for decentralized and consensus consistency. But these features may result in weak performance.
In the case of bitcoin, a block is collection of about 3,000 transactions. When a block is verified, all 3,000 transactions are validated. If not encapsulated, transactions have to be validated one by one, which means high-frequency validation requests and low efficiency. Meanwhile, each block contains addresses of the previous block to form a “chain” from the latest block all the way back to the original genesis block. New blocks can be created in many ways and forks can happen. For bitcoin, the rule “selecting the longest chain as the main chain” helps identify the recognized “main chain” after a certain time even if forks happen.
This rule is one of the reasons behind low throughput.
EOS, ThunderChain and Ethereum have all turned to the multi-chain architecture to address the performance issues, even though they got different in name.
Apparently, single-machine computing, memory and disk resources are limited. An ideal blockchain architecture should support a larger number of machines effectively and allows performance scalability by adding new machines.
The architectures adopted by EOS, ThunderChain and Ethereum are different. EOS uses a 21-hypernode environment, which means high competence requirements for these hypernodes in terms of storage, bandwidth and computing power.
In accordance with EOS’s hardware specifications, the server configuration scheme for node candidates should meet these minimum requirements: Amazon AWSEC 2 host x1.32x large, 128-core processor, 2TB memory, 2x1920GB SSD, and 25Gb bandwidth.
ThunderChain has advantages in this aspect. It has deployed a mega-node shared computing nodes (OneThing Cloud). These nodes serve as bookkeeping nodes and offer high computing power as a whole under a well-designed architecture to support millions of transactions concurrently.
EOS has created a mechanism encouraging voluntarily spending on high-performance blockchain environments. EOSYS, an EOS Block Producer (BP) candidate, has estimated maintenance costs to exceed US$90,000 each month, or US$1.08 million annually.
Which is the best?
In terms of application scenarios, all three blockchains have their own advantages and limitations. For example, ThunderChain doesn’t support ICO projects while Ethereum and EOS can do so.
In terms of R&D, ThunderChain and Ethereum are developed based on Solidity. EOS is using the programming language C++ and development tools are limited so far.
However, EOS developers do not need to spend transaction fees, though they have to pay for EOS storage and computing resources.
ThunderChain and Ethereum have transaction fees. ThunderChain has promised to remove transaction fees for the first year (May 16, 2018 – May 16, 2019) to further reduce development cost. Ethereum charges an average fee rate of US$ 0.05 per contract.
With 35 million users, Ethereum remains the most popular blockchain platform among others. Currently, EOS has only 230,000 users. ThunderChain is built upon over 1 million Xunlei’s blockchain devices and Xunlei has over 400 million users in total.