Proposal to allow ANOs to go down to one server if so desired

Public: Only invited members may reply

  • Viewed BI Foundation BI Foundation Bedrock Solutions Bedrock Solutions Blockrock Mining Blockrock Mining Canonical Ledgers Canonical Ledgers Consensus Networks Consensus Networks Cube3 Cube3 David Chapman De Facto De Facto Factable Solutions Factable Solutions Factom Inc. Factom Inc. Factomatic Factomatic Factomize Factomize Factoshi Factoshi Federate This Federate This Go Immutable HashQuark HashnStore HashnStore Kompendium Kompendium LUCIAP LUCIAP LayerTech LayerTech Matters Matters Prestige IT Prestige IT RewardChain RewardChain Stamp-IT Stamp-IT The Factoid Authority The Factoid Authority VBIF VBIF
  • Not Viewed Crypto Logic Crypto Logic DBGrow DBGrow Multicoin Capital Multicoin Capital

Can ANOs choose to go down to one Authority Node and not go back to two?


Have not voted

Authority Nodes BI Foundation BI Foundation Blockrock Mining Blockrock Mining DBGrow DBGrow VBIF VBIF

  • Total voters
    23
  • Poll closed .

David Chapman

Factomize
We need to act proactively and allow ANOs to implement cost saving measures that will allow the Factom Protocol to weather the current health and economic issues faced by the world. I propose that ANOs be allowed to go down to one Authority Node if 60% of ANOs vote to pass this measure. They can then adjust their efficiency for their single server as needed. This measure would have immediate effect and the ANO cannot go back to two servers in the future.

This is not a Major Discussion as that requires inviting Guides. I will create a poll for ANOs after 8 days (on the 22nd) of discussion as if this were a Major Discussion.

Thank you.
 
Last edited:

Alex

Factoshi
I think this is a good idea. We don't gain much in the way of decentralisation by requiring each ANO to run two servers. The only reservation I have is how it may impact protocol stability. Honestly, I am not too concerned about that, though. I believe any drawback would likely be marginal and is outweighed by the flexibility it offers ANOs during what is likely to be a very trying time for us all.

Once an ANO reduces to a single server, would they be able to reclaim a second server in the future, or do you intend for it to be a one way street?
 

Alex

Factoshi
Hopefully anyone dropping to 1 server and 0% efficiency will not see it's standing removed because it is lower than 10% :D
This is actually a good point. It may be wise to standardise efficiency so that it can be compared across ANOs without knowing how many servers each ANO is running.

Perhaps an ANO with only one server at 0% efficiency should still be considered a 50% ANO? Or perhaps we should compare FCT renumeration directly? If neither of those are viable, I think we should add server count to the standing dashboard.
 

David Kuiper

Bedrock Solutions
The only reservation I have is how it may impact protocol stability. Honestly, I am not too concerned about that, though. I believe any drawback would likely be marginal and is outweighed by the flexibility it offers ANOs during what is likely to be a very trying time for us all.
FYI, @Steven Masley gave his thoughts about how this might affect the network in the previous discussion on this topic.

We should also note that @BI Foundation already made the decision to go to one server in a previous update.

I support allowing ANOs to voluntarily go down to one server. There is no need to waste resources on extra servers if they have a minimal or possibly negative impact on the protocol.

and the ANO cannot go back to two servers in the future.
I'm not against this necessarily, but I'd like to hear the reasoning.
 

Miguel Proulx

Stamp-IT
As long as others ANOs have the possibility to drop down below 50% for special secret projects that might help the protocol instead of going the grant route, I think we should also allow ANO who decided to drop to 1 server, to get their server back if they have a project that would require the extra funds to see the light. Just my 2c.
 

Michael Lam

Core Developer
As long as others ANOs have the possibility to drop down below 50% for special secret projects that might help the protocol instead of going the grant route, I think we should also allow ANO who decided to drop to 1 server, to get their server back if they have a project that would require the extra funds to see the light. Just my 2c.
I think thats too complicated. Effectively any ANO dropping one server should know that the most revenue they can stream post drop would be a functionaly equivalent of 50% efficiency on two servers (0% efficiency on 1). Of course grants would be available as usual to cover any additional projects. A reversal back to two servers would be allowed if they get proper agreement from standing parties to go back up, but that may not be granted, and would potentially take a while to accomplish.

Thats the way I would interpret this proposal. Vote how you like. I will probably vote for this. If an ANO would like to drop to 1 server (and nobody forces them to), why shouldn't they be allowed to? For instance, CryptoLogic operating at 70% efficiency on two servers, could operate 1 at 40% efficiency and pocket the savings. For some ANOs this would make sense.
 

Keith Pincombe

Marketing Working Group
Governance Working Group
I'm not really against David's original suggestion of every ANO going down to one server. I think it should be reconsidered. There is any technical benefit to running two servers. The only negative is that it limits the maximum an ANO can earn.

Considering the current changes to governance, grants, standing and the current value of FCT. I personally think it's difficult to argue that the FCT is necessary to operate.
 

Paul Bernier

LUCIAP
Core Committee
Core Developer
The one way street part bothers me a bit, as it would create 2 separate classes of ANOs, irremediably and regardless of future conditions (or theoretically until we have 65 ANOs, which is anyone guess when that would actually happen). Having non equal ANOs would change quite a bit the dynamic of our governance (first class citizens, which can modulate their efficiency with bigger amplitude, and second class citizens who can't). I'd like Luciap to save some bucks right now by going to 1 server, but would be reluctant to do so by fear of becoming a second class citizen for ever.
 
Last edited:

Jason Gregoire

Kompendium
This option already exist today and requires no decree: any ANO who wishes to only operate one Auth server has every ability to do so.

This initiative IMO is pure politicking and just serves to socially pressure those who choose to run two servers to capitulate to those who want everyone to run one.
 

Who

Factomize
Core Committee
Core Developer
We don't gain much in the way of decentralisation by requiring each ANO to run two servers. The only reservation I have is how it may impact protocol stability.
The consensus mechanism requires at least some Audit nodes to use as backup when a Leader fails. At the moment, there are 27 ANOs, 28 leaders and 25 audits. There is a lot of room for some ANOs to go down to one server but it becomes a problem if too many do. Having 28 leaders and 0 audits is a huge network stability liability, so if the number of ANOs going down to one server drops too low, the number of leaders would have to be reduced as well. That decreases decentralization.

I'm not sure what the best ratio of L:A is but I wouldn't feel good about having less than a 2:1 ratio. There are potential situations like amazon outages that could knock out a big chunk of the servers and without enough redundancy the network would be paused during that time.

tldr: going down to one server is okay if only some ANOs do it, but if too many ANOs do it, it starts to affect both decentralization and stability.
 
The 28 leaders is a hardcoded value that could be reduced as needed though? It would be a clunky way to do it every update but not necessarily impossible. ANO's could announce in advance and then decommission when the next version is released.
 

Who

Factomize
Core Committee
Core Developer
The 28 leaders is a hardcoded value that could be reduced as needed though?
It's regulated via the admin chain and can be updated on the fly (ie the onboarding/offboarding process). It can be reduced if necessary but doing so would decrease decentralization.
 

Alex

Factoshi
Can you expand a little on the decentralisation risk, @Who? I'm not entirely convinced that reducing the absolute number of leaders meaningfully worsens decentralisation. There are a variety of reasons for that, including the fact that we would not reduce the number of ANOs and that we are all connected to the Docker Swarm anyway. However, I recognise you may have some special insight into this, so I am keen to hear your view on it.
 

Who

Factomize
Core Committee
Core Developer
Can you expand a little on the decentralisation risk, @Who? I'm not entirely convinced that reducing the absolute number of leaders meaningfully worsens decentralisation.
"Meaningfully" is the key qualifier here. Proof of authority provides the immutability of all the data in the chain. The fewer ANOs involved in ensuring the integrity, the less decentralization there is objectively. If an ANO resigns it's not a big deal, the same way if a handful of ANOs reduce their server count to one.

What constitutes a meaningful decrease of decentralization is up for debate.
 

Alex

Factoshi
The fewer ANOs involved in ensuring the integrity, the less decentralization there is objectively.
To be clear though, we would not be reducing ANOs, only the authority nodes themselves. This is why I am not really convinced we would be much worse off. I can see some risk to decentralisation in that ANOs with two nodes would have a louder voice (in terms of what constitutes truth) than those with only one.
 

Who

Factomize
Core Committee
Core Developer
To be clear though, we would not be reducing ANOs, only the authority nodes themselves.
From a cryptographic standpoint this doesn't make a difference. Audit servers don't sign the blocks, so if there are fewer leaders than ANOs, it's theoretically impossible for all ANOs to sign the blocks.

From a practical standpoint, you're right. Due to server outages/elections, there are currently some ANOs without a leader already. If we lowered the number of leaders it wouldn't change much from how it is right now, other than lowering the theoretical best case scenario.
 

Alex

Factoshi
The one way street part bothers me a bit, as it would create 2 separate classes of ANOs, irremediably and regardless of future conditions (or theoretically until we have 65 ANOs, which is anyone guess when that would actually happen). Having non equal ANOs would change quite a bit the dynamic of our governance (first class citizens, which can modulate their efficiency with bigger amplitude, and second class citizens who can't). I'd like Luciap to save some bucks right now by going to 1 server, but would be reluctant to do so by fear of becoming a second class citizen for ever.
I am worried about this too. I am left wondering whether Factoshi would be better to go to one server or to lower efficiency in order to improve cost to income ratio. The former is more cost effective for the protocol, but if it’s a one way street then the latter is a safer idea for Factoshi as a business as it retains more options.

Can we make this a multi option vote so that ANOs can express their preference?
 

Colin Campbell

Federate This
Marketing Working Group
I think it’s a great initiative to reduce server costs as much as possible. I would like to retain the ability to go back to two, or maybe through standing.

If your standing is above e.g 80% you can reclaim your 2nd server.
 

Beta@HashQuark

HashQuark
HashQuark support this initiative. However, two aspects that we need to take into consideration. One is the side effect or potential risks if too many ANOs drop to 1 server. Another is more choices for ANOs. For example, ANOs could choose to reduce the efficiency once they drop to 1 server.
 

Saul Schwartzbach

Prestige IT
Since some ANOs have already publicly announced to moving to one server prior to this initiative, is the only real change being proposed here is the fact that this decision can't be reversed?

My main question would be, are those who were already planning to drop a server reconsidering now that this may become a one way street?
 

Alistair McLeay

RewardChain
Because it is voluntary, we will be voting in support. We would prefer the option to go back to two remained open, though—at least until we have 32 ANO's on board. With our current governance situation we believe we should not be going above 32 ANO's for quite some time, so keeping the option to move back to 2 makes sense.
 
Top