[
https://issues.apache.org/jira/browse/IGNITE-24064?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Denis Chudov updated IGNITE-24064:
----------------------------------
Description:
*Motivation*
The RendezvousDistributionFunction should be able to distribute learners.
Also, the additional requirement here is that consensus group replicas should
be distributed by cluster as evenly as the algorithm allows for all replicas.
This is required because the consensus group bears additional load related to
providing the data consistency; also, there are group leaders which are chosen
from the members of the consensus group.
*Definition of done*
The RendezvousDistributionFunction is able to distribute learners and
distributes the consensus replicas (voters) as even as it possible for this
type of algorithm.
*Implementation notes*
*The assignments calculation for learners via Rendezvous should be simple:
Rendezvous algorithm builds the list of nodes for each partition, then takes
first N of them (where N is replica factor) as assignments. We should modify it
so that it would take the first N nodes as a consensus group (where N is the
size of this group), and the next M nodes as learners (where M is learners
count).*
> Improve the RendezvousDistributionFunction in order to calculate the
> assignments of peers and learners
> ------------------------------------------------------------------------------------------------------
>
> Key: IGNITE-24064
> URL: https://issues.apache.org/jira/browse/IGNITE-24064
> Project: Ignite
> Issue Type: Improvement
> Reporter: Denis Chudov
> Priority: Major
> Labels: ignite-3
>
> *Motivation*
> The RendezvousDistributionFunction should be able to distribute learners.
> Also, the additional requirement here is that consensus group replicas should
> be distributed by cluster as evenly as the algorithm allows for all replicas.
> This is required because the consensus group bears additional load related to
> providing the data consistency; also, there are group leaders which are
> chosen from the members of the consensus group.
> *Definition of done*
> The RendezvousDistributionFunction is able to distribute learners and
> distributes the consensus replicas (voters) as even as it possible for this
> type of algorithm.
> *Implementation notes*
> *The assignments calculation for learners via Rendezvous should be simple:
> Rendezvous algorithm builds the list of nodes for each partition, then takes
> first N of them (where N is replica factor) as assignments. We should modify
> it so that it would take the first N nodes as a consensus group (where N is
> the size of this group), and the next M nodes as learners (where M is
> learners count).*
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
