[ 
https://issues.apache.org/jira/browse/HELIX-674?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16409890#comment-16409890
 ] 

ASF GitHub Bot commented on HELIX-674:
--------------------------------------

Github user mcvsubbu commented on a diff in the pull request:

    https://github.com/apache/helix/pull/145#discussion_r176490613
  
    --- Diff: 
helix-core/src/main/java/org/apache/helix/api/rebalancer/constraint/dataprovider/CapacityProvider.java
 ---
    @@ -0,0 +1,48 @@
    +package org.apache.helix.api.rebalancer.constraint.dataprovider;
    +
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + *   http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing,
    + * software distributed under the License is distributed on an
    + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
    + * KIND, either express or implied.  See the License for the
    + * specific language governing permissions and limitations
    + * under the License.
    + */
    +
    +/**
    + * An interface for getting participant capacity information.
    + * The return value will be used to estimate available capacity, as well 
as utilization in percentage for prioritizing participants.
    + *
    + * Note that all return values of the provider are supposed to be in the 
same unit.
    + * For example, if the provider is for memory capacity of a host (total 
memory size is 8G, current usage is 512MB),
    + * the return values could be {@link 
org.apache.helix.api.rebalancer.constraint.dataprovider.CapacityProvider#getParticipantCapacity(String)}
 = 8192, and {@link 
org.apache.helix.api.rebalancer.constraint.dataprovider.CapacityProvider#getParticipantUsage(String)}
 = 512.
    + * Another example, if the provider is for partition count capacity (max 
count 1000 partitions, currently no partition assigned),
    + * the return values should be {@link 
org.apache.helix.api.rebalancer.constraint.dataprovider.CapacityProvider#getParticipantCapacity(String)}
 = 1000, and {@link 
org.apache.helix.api.rebalancer.constraint.dataprovider.CapacityProvider#getParticipantUsage(String)}
 = 0.
    + *
    + * Moreover, while this provider is used together with a {@link 
org.apache.helix.api.rebalancer.constraint.dataprovider.PartitionWeightProvider 
PartitionWeightProvider},
    + * both providers are supposed to return values in the same unit so they 
can be used to estimate the resource usage of an proposed assignment.
    + */
    +public interface CapacityProvider {
    +
    +  /**
    +   * @param participant
    +   * @return The total participant capacity.
    +   */
    +  int getParticipantCapacity(String participant);
    +
    +  /**
    +   * @param participant
    +   * @return The participant usage (provisioned capacity).
    --- End diff --
    
    The comment "(provisioned capacity)" is confusing. Can we remove it?


> Constraint Based Resource Rebalancer
> ------------------------------------
>
>                 Key: HELIX-674
>                 URL: https://issues.apache.org/jira/browse/HELIX-674
>             Project: Apache Helix
>          Issue Type: New Feature
>            Reporter: Jiajun Wang
>            Assignee: Jiajun Wang
>            Priority: Major
>             Fix For: 0.8.x
>
>         Attachments: Constraint-BasedResourceRebalancing-080318-2226-240.pdf
>
>
> Helix rebalancer assigns resources according to different strategies. 
> Recently, we optimize the strategy for evenness and minimize movement. 
> However, the evenness here only applies to partition numbers. Moreover, we've 
> got more requests for customizable rebalancer from our users.
> Take partition weight as an example:
> In reality, partition replicas have different size. We use "partition weight" 
> as an abstraction of the partition size. It can be network traffic usage, 
> disk usage, or any other combined factors.
> Given each partition may have different weights, Helix should be able to 
> assign partition accordingly. So that the distribution would be even 
> regarding the weight.
> In this project, we are planning new rebalancer mechanism that generates 
> resource partition assignment according to a list of "constraints". Current 
> rebalance strategy can be regarded as one kind of constraint. Moving forward, 
> Helix users would be able to extend the constraint interface using their own 
> logic.
> Some init discussions are in progress and we will have a proposal posted here 
> soon.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to