Wed. Jul 28th, 2021
    Rust Cup, Free TON,

    It has been almost six months since the participants of the Free TON Rust Cup race took their starting positions to test the network. And so, on July 6, the validator contest on the new Rust node finally started. FTH magazine asked the participants to share their impressions of the months of waiting and the beginning of the contest.

    Rust Cup is the third large-scale Free TON validator contest after Magister Ludi and DePool Game. Attention to the Rust Cup was especially close because of the ambitious goals of the Cup and the solid value of the prize pool — almost 60 million TON Crystal.

    First, the race should bring a network built on nodes written in the Rust programming language into working conditions. In a way, this is the creation of the entire network anew, after which it should become faster than all currently existing in the crypto industry. Along the way, the Rust Cup sets the goal of setting a world record for the speed of smart contracts execution.

    Second, as part of the race, the system of punishing validators whose work may threaten the stability of the network is being tested — slashing. We talked about why this is a critical element of the Free TON blockchain.

    A Tense Wait

    All the validators we talked to — Anatoly Ustinov, Dmitry Gachko, as well as representatives of the FreeTON.One project — Dmitriy M (CEO) and Axel F (CTO) especially emphasized the duration of the pre-launch period.

    Dmitry Gachko called this condition — “Burnout by waiting”.

    At the same time, the validators understand that during this period, the development team has done a titanic job of improving the network software. 

    Dmitry M and Axel F noted that remembering the launch of Magister Ludi, they were prepared for a protracted start of the Rusty Cup and expected a delay of a quarter to two. As a result, they guessed the right edge of the range.

    Since February, we have experienced a lack of correct information about the start of the race. Both we and potential participants incurred tangible costs, not knowing when the start would be. Community engagement and an engineering sense of humor helped us get through this period. And the rewards for early testing will be a nice bonus. Dmitriy M

    At Idle Speed

    Forced downtime was also used by validators to choose the most suitable hardware configuration. Anatoly Ustinov noted that the impressive costs were required not only for the hardware but also for the validation software development, which his team can now be proud of.

    The FreeTON.One project leaders admitted that, with Magister Ludi’s experience, it is possible to claim victory with equipment that significantly exceeds the parameters recommended by the conditions of the contest. The team decided to increase the chances of winning by using more than one machine for the tests. Also, the validators noted that this time they decided to experiment by placing capacities in data centers located in different geographic zones.

    Rust Cup: Ready! Steady! Go!

    From Dmitriy Gachko’s point of view, the start looked rather unexpected, because there was no transition to the standard validation cycles — the contest rules say about 37 laps, which implies standard 18-hour cycles. In reality, the validation cycles were shortened, which, in his opinion, was the reason for the quick failure of the new network.

    Dmitriy M and Axel F said that the announcement of the start caused ambivalent feelings. On the one hand, they participated in the tests from the very beginning and were ready since February. In addition, it was discussed at the Free TON weekly meetings that the latest updates of the network and the node signal a “warm-up” and the race will begin soon.

    On the other hand, comparing their experience with the Magister Ludi contest, they believe that, in some ways, the beginning did take them by surprise. For example, the race schedule was published only two-and-a-half hours after the start.

    The announcement of the start time of the race at the very start caused, at least, surprising. Dmitriy M

    Anatoly Ustinov, however, noted that the launch announcement was previously stated by the co-founder of TON Labs, Alexander Filatov, during an interview for the Ghost In the Block telegram channel.

    First Failure

    A few days after the start of the Rust Cup, the network went down. This is the flip side of blockchain scaling and the original Free TON slashing mechanism.

    Recall that the blockchain is automatically divided into shards, which are provided by the work of the validator nodes. The adopted slashing mechanism can also automatically remove nodes from the network that do not meet the accepted correctness criteria. A critical number of blocked nodes lead to a crash of the entire network. This happened several times during the testing of the Rust Cup, and it happened soon after the start.

    The crash came as little surprise to anyone — the launch of the network in a stressful mode was planned — with reboots, node failures, and changes in the protocol during the race.

    Preliminary Conclusions

    As it turned out, the race participants were ready for many challenges over the months of waiting and with the experience of previous contests.

    Anatoly Ustinov looks at the challenges ahead, and believes that “it’s even more interesting this way”. He notes the positive changes that have occurred during the test launch of the network.

    The node has been significantly redesigned, compression has been added, traffic has been significantly reduced, and the load on the processors has also been reduced. Anatoly Ustinov

    Dmitry Gachko speaks ironically about the problem areas of the Rust network.

    The organizers are more likely to have difficulties, especially in the issue of counting places: with a network restart and a shard downtime, there may be different opinions and results. Dmitry Gachko

    Dmitriy is concerned only about the communication channels since test launches showed a bottleneck in them.

    Axel F spoke about a pleasant surprise as an updated RUST node, which, in comparison with earlier versions, can take part in processing 3,000 transactions per second with almost four times less utilization of system resources.

    It was nice to see ~150 megabits on the network port at such speeds. I can’t wait to see what the network and we, as validators, are capable of. I would like to wish the organizers of the contest to achieve the same success as the RUST node development team. Axel F

    Rust Cup: Waiting For The Record

    “Will there be a world record for the number of transactions per second?” — we addressed this question to the contest participants.

    Dmitry M and Axel F believe that there are still many surprises and difficulties ahead of the validators, so they have not yet formed an unambiguous opinion — it is difficult to imagine how the network will behave when trying to set a record.

    There is a risk that during the achievement of record values, the Internet bandwidth and the loyalty of providers in data centers will become a bottleneck. Our common task is to understand where and when we will reach the limit. Axel F

    As a reminder, the record requires crossing the threshold of 50,000.

    Anatoly Ustinov expressed confidence in the possibility of setting a record. Although this requires that the network has enough nodes with good hardware. According to his calculations, at the moment there are just over 100 out of over 300. The second important point that he drew attention to was the need for reliable dApp servers for the contest to keep the ability to submit during the election.

    17
    1