Niall Yorke

The Refiner Incentivized Testnet Recap

On July 26th, we concluded an amazing two-week Refiner Incentivised Testnet. And following its conclusion, the team has great confidence in not only the performance of the Refiner on mainet, but the network operators joining the CQT Network. We would like to give a huge thanks to all who participated and helped us make a better CQT Network.

When we decided to run a Refiner Testnet, there were a number of goals we wished to achieve, namely:

  • Test the total throughput of the network with both block specimens and block results being produced,

  • Gather user feedback and gauge how easy or difficult it is to run the Refiner,

  • Identify and select a new cohort of operators to run the Refiner on mainnet,

  • Check for the usability of the Refiner stack tooling and the readability of the documentation, and

  • Test the front end at scale for onboarding new roles to the CQT Network.

We are delighted to say we achieved all the above.

Testnet by Numbers and Network Throughput

Over the course of the two weeks:

  • 27 individuals and organizations ran the Refiner

  • 274,833 proofs were submitted between July 12th and July 26th

  • An average uptime of 98.8% was achieved

To accomplish testing the network throughput with the Refiner live, a systematic approach was taken to progressively push the limits of the network's performance, allowing for an in-depth understanding of its capabilities.

At the outset of the testnet, the focus was on initializing the source block specimen event listener and running the Refiner transformation pipeline. Consequently, the goal was to increase the number of source block specimens generated and finalized so that more data objects could be picked up by the Refiner pipeline. This process was carefully managed to maintain stable conditions within the network throughout. The intention behind controlled increments to block specimens produced was to ascertain if the current network parameters were possibly constraining the potential production capacity of block results per second. The answer to this was clearly that the network could manage the consistent increases without major glitches or any kind of stalling. A few 10s of timeouts in uploading the block results were noticed in operators due to latencies observed in the responses from the third-party IPFS network interfaces.

The testnet participants played a pivotal role in this effort. At the commencement of the testnet, one block result was produced approximately every ~7 minutes. This baseline was gradually augmented, and significant progress was observed. By the penultimate day of testing, a block result was being generated roughly every four minutes. The momentum was sustained, leading to a remarkable achievement on the final day of the testnet, wherein the network was effectively producing ~ 2/3 block results in approximately every minute. As seen below, participants were submitting more proofs over the course of the two weeks as we increased the parameter determining how many block results were produced.

It's important to note that this acceleration in block results production was accompanied by a noticeable rise in gas consumption. Consequently, careful attention was required to ensure that the operators had an adequate gas supply to sustain the heightened activity. Additionally, there was a controlled and manageable increase in the load requirements imposed on Refiner node servers. This meticulous approach was essential in evaluating the optimal performance states of both the Refiner stack and the Network Operators, thus contributing to the overall determination of network parameters for the Refiner mainnet.

This stress testing identified two notable bottlenecks:

  1. Overall gas consumption

  2. As the block result production rate was increased and more block specimens were fetched, the Filecoin network could not keep up with the request rate.

The CQT Network team is evaluating mitigating the first with solutions, including developing our own L1. To address the second bottleneck, we are actively working towards developing a better fail-proof fetching and uploading mechanism to subvert any such issues in the future.

Refiner Feedback

The Refiner Incentivised Testnet not only provided a platform to assess technical capabilities but also garnered valuable feedback from participants. The responses highlighted the remarkable balance achieved between the Refiner's ease of use and its robust performance.

One participant, after actively participating in the testnet for 11 days, praised the seamless operation of the Refiner.

Another contributor applauded the Refiner's installation and configuration process as well as the tool's data transmission speed and efficiency.

In a comprehensive reflection, another participant shared a positive overall experience with a few constructive suggestions. They praised the clear and user-friendly instructions and highlighted the need for a broader compatibility with different versions of Ubuntu. Technical challenges, including port scanning issues and temporary restrictions, were acknowledged. Despite these hurdles, the participant found that the Refiner and the Testnet largely met their expectations, recommending it to others.

New Network Operators Joining CQT Network

As the Refiner Incentivised Testnet reached its conclusion, it's with great pleasure that we recognize the exceptional participants who have been selected to join Mainnet as winners. These individuals have exhibited outstanding dedication and prowess, aligning perfectly with the vision of a resilient and high-performance CQT Network.

The selection process was comprehensive, taking into account multiple facets to ensure that the chosen winners possess the qualities required to elevate the network. The assessment criteria included:

  1. Proofs Submitted: The ability to sustain a live presence was evaluated through the number of proofs submitted during the testnet period. This criterion exemplifies the operational robustness and reliability demonstrated by each participant.

  2. Experience in Infrastructure: Prior experience in running infrastructure, as indicated in the application form, was a significant factor. This experience underscores the participants' proficiency and familiarity with the technical intricacies of network operations.

  3. Overall Participation: Active engagement, encompassing activities such as asking questions, providing support, and contributing insights, was a noteworthy factor. It exemplifies a collaborative spirit and a commitment to the network's collective growth.

The winners selected for mainnet entry bring a wealth of experience to the CQT Network. Their contributions during the Refiner Incentivised Testnet have not only demonstrated their technical capabilities but also showcased their dedication to enhancing the network's performance and resilience.

What's Next?

With the Refiner Incentivised Testnet now behind us, a new chapter awaits as the Refiner gears up for its debut on Mainnet, scheduled for August 15th.

The achievements of the Refiner Incentivised Testnet serve as a testament to the power of collaboration among our participants. Their engagement and unwavering dedication have subjected the network to rigorous testing, ensuring a poised and efficient launch of the Refiner on the mainnet.

The Covalent team extends its gratitude to all who have contributed, and we look forward to the CQT Network's continued evolution and success.

About Covalent:

Covalent provides the industry-leading Unified API bringing visibility to billions of Web3 data points. Developers and analysts use Covalent to build exciting multi-chain applications like crypto wallets, NFT galleries, and investor dashboard tools utilizing data from 225 + blockchains. Covalent is trusted by a community of 40,000+ developers and powers data for 5,000+ applications, including 0x, Zerion, Rainbow Wallet, Rotki, Bitski, and many others.