When Do Transactions Fail?

    Do times of high transaction volume lead to more transaction failures? Create a visualization that shows the relationship between transaction volume and success rate of transactions since May 9th. Are there any noticeable patterns to when transactions are more likely to fail? Were there specific events that seem to cause transaction failures?

    Introduction and methods

    Flow blockchain is a fast, decentralised, developer-friendly PoS-based blockchain from NFT pioneer Dapper Labs that aims to provide high scalability to the next generation of Dapps without the use of complex scaling techniques such as fragmentation.

    Flow applies the pipelining technique by separating the jobs of a validator node into four different roles: collection, consensus, execution and verification. For each role, a minimum participation deposit is required from each of the participating nodes. All these nodes work in parallel to perform their required actions. Each validating node still participates in the validation of each transaction, but does so only in one of the validation stages.

    Over the past months, the volume and activity on Flow seems to be increased and it was more prominent after the UST depegging event on Terra. The idea of this dashboard is to try to assess if the more transactions volume lead to more transactions failure.

    To do that, we will analyze the following metrics:

    • Daily transactions by status
    • Transactions volume and succeeded rate over time
    • Succeeded vs failed transactions rate over time
    • Comparison between transactions volume and succeed rate

    Results

    Loading...
    Loading...

    The first analyzed chart represents the daily transactions by status (failed or succeeded). In this case, we can see how the number of transactions increased from May 11th to May 16th during the UST depegging period. However, afterthat, it returned progressivelly to previous values. After a period of consolidation around 4M daily transactions, the number started to decrease in May 25th not only in terms of transactions but only in percentage of succeeded transactions. We can see how during the last week, the amount of failed transactions increased dramatically while the total transactions decreased.

    Taking a look at the second chart about the transactions rates comparison, we can see how during the major of the period, the succeeded rate was around 0.95, however during the last days, the rate started to decreased reaching less than 0.30. Then, the succeeded rate is being surpassed by the failed rate, representing that currently there are more failed transactions than succeeded transactions on FLOW.

    Loading...

    In these final charts, we can see several comparisons between the volume on FLOW and the succeeded transactions rate. The first image shows the daily numbers, in which we can se how the rate remained constant during the first large period while the volume had a waves behaviour. However, when the major peak of volume was detected, the succeeded rate started to dump. Afterthat, even the volume went up or down, the succeeded rate continued dumping. In the final picture, we can see how there is not a significant correlation between these metrics.

    Loading...

    Conclusions and key insights

    In this dashboard we have analyzed the failed transactions evolution on Flow as well as the possible causes. The main assumptions are:

    • The amount of failed transactions increased dramatically while the total transactions decreased, coinsiding with the highest volume registered.
    • During the major of the period, the succeeded rate was around 0.95, however during the last days, the rate started to decreased reaching less than 0.30.
    • There is not a significant correlation between success rate and volume.
    • The basic events taht caused this increase on the number of failed transactions have been the FlowToken and FlowFees contracts.
    Loading...
    Loading...
    Loading...

    In this final charts we can see how the main events that are causing this decrease on the succeeded transactions rate are the FlowToken and FlowFees. 66.7% of the failed transactions come from the first one and the rest for the secodn one. The rest of the contracts didn't show any failed transactions.

    We can see how the major of the failed transactions have been registered during the last days when the failed transactions dramatically increased, dumping the success rate of both type of events from around 0.95 to below 0.2!