Performance Based Node Rewards

Following up on the discussion in the November Node Provider Working Group, we have been analyzing how we can address the impact of high subnet load and protocol changes on node performance. Specifically, based on the discussion in the meeting, we have explored how to differentiate systematic failure rates (affecting all nodes in a subnet) from idiosyncratic failure rates (specific to individual nodes).

Subnet Failure Rate Analysis

Since April 2024, subnet failure rates (FR) have been analyzed.

  • We compare the median and 75% quantile of the daily failure rate of nodes within the subnets w4rem, fuqsr for the time period April to November 2024.
  • Most of the time these measures are at very low levels, i.e., below 2%.
  • In October, we observed a systematic increase in the subnet failure rate for all investigated subnets. This is reflected in an increased median and 75th quantile.


Suggested Methodology for the Determination of the Node Failure Rate

Following the discussions in the Node Provider Working Group and the analysis presented, it is recommended distinguishing between systematic and idiosyncratic node failure rates. We propose that only the idiosyncratic component of node failure rates should influence reward multipliers. This means that a node would be penalized only if its performance significantly deviates from the performance of its peer nodes within the same subnet. This approach can be detailed as follows:

Systematic Failure Rate

  • Calculate the 75th percentile failure rate daily for each subnet to account for systematic factors such as protocol changes or high load of subnet.
  • This provides a mapping: DAY -> SUBNET_ID -> SYSTEMATIC_FR

Idiosyncratic Failure Rate

  • To isolate the idiosyncratic failure rate for a node, compute the difference between a node’s daily failure rate and the subnet’s systematic failure rate.Apply flooring to avoid negative values: Idiosyncratic Failure Rate = max(0, Node Failure Rate - Systematic Failure Rate)
  • Only the idiosyncratic failure rate is then used as an input for the calculation of the reward multiplier.

Example Calculation:

Example 1:

This example shows a node on the k44fs subnet whose rewards would have been adjusted in the prior approach considering the absolute node failure rates. However, since the failure rate is systematic, no adjustment was applied.

Considering both systematic and idiosyncratic components the failure rate is 12.89% which corresponds to a rewards multiplier of 96.6%.

Rewards XDR calculation:

  • Base monthly rewards XDR: 1584
  • The idiosyncratic failure rate for a node in a reward period is computed averaging the daily idiosyncratic failure rates.
    In this example the idiosyncratic failure rate is 3.02%. The node will be rewarded fully without adjustments.

Example 2:

This example shows a node on the w4rem subnet whose rewards are adjusted significantly since the failure rate is idiosyncratic, i.e., exceeding the 75% quantile of the failure rates of the subnet.

Considering both systematic and idiosyncratic components the failure rate is 36.02% which corresponds to a rewards multiplier of 70.2%.

Rewards XDR calculation:

  • Base monthly rewards XDR: 2157.25
  • The idiosyncratic failure rate is 35.14% which corresponds to a rewards multiplier of 71.2%.
  • In this example, the systematic component is minimal, as most of the other nodes in the subnet have performed well during the period.
  • The node is rewarded 2157.25 * 71.2% = 1536 XDR.
8 Likes