Uptime Requirement

#1

This is part of a set of threads for focused discussion on specific aspects of the 3.50 proposal (formerly known as 3.12).

To get up to speed on 3.50, see the pinned post in this category for links to more information.

Uptime and Farming Rewards

While a minimum uptime requirement for nodes to mint TFT has been in our specification since before v3, it has not yet been enforced. Currently, nodes mint proportionally to the uptime that they achieve.

The 3.50 release is a good opportunity to put this system into place.

Review

Our published uptime requirements have so far depended on the certification level of the node:

  • DIY nodes - 95%
  • Certified nodes - 98%
  • Gold certified nodes - 99.8%

This assumes that the nodes are in regions with stable infrastructure. For locations with less stable infrastructure, it’s proposed that the requirements can be lower. However, this might be too complicated and difficult to apply in a fair way.

Idea

With a reduced base reward in 3.50 and more of farmers revenue coming from utilization, we can consider a different approach to uptime requirements. Since nodes with low uptime will naturally receive less utilization and thus less revenue to the farmer, based on market forces, we can apply a blanket uptime requirement that is appropriate for all regions (for example, 70%).

There has also been some discussion of a more complex model where uptime levels have varying impacts on the farmers rewards. Read more about that here.

#2

Uptime and Farming Rewards: Blanket Uptime Requirement Model

  • < 70% uptime: no farming rewards
  • >= 70% uptime: farming rewards proportional to uptime
#3

We also need to think about how people will perceive our uptime requirements, regardless of what is actually happening on the grid. I’m worried with a 70% requirement people will attack that as being a very low uptime.

1 Like
#4

There’s a few ways to think about it. One is the minimum uptime for a node to be able to perform some subset of useful tasks. At 70%, a node could provide archival storage, would work fine as a QSFS backend, and would even be useful as one of several redundant pieces in a deployment requiring greater liveness. With a market based system, lower uptime nodes would have less demand and thus the farmers running those nodes would not be able to demand a markup for their capacity. Users who want a lower cost and don’t need the uptime for their use case could take advantage of this.

Another way is considering the Grid as a global network and setting minimum that can work for a reasonable percentage of regions. For example, in South Africa, there are “load shedding” planned power disruptions that typically cover at least 12% of time in a month. A 95% requirement just doesn’t work in these areas. Implementing a set of pre region requirements is a big can of worms, and I think we should seriously consider any viable alternative.

I get what you’re saying here, but the facts are pretty compelling. Based on some minting data from one month earlier this year:

  • About 75% of nodes achieved greater than 95% uptime
  • About 8.5% of nodes achieved between 70-95%

This means that with a 70% requirement we still exclude 2/3 of the nodes that don’t meet the 95% threshold.