[
https://issues.apache.org/jira/browse/HBASE-13965?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14621457#comment-14621457
]
Lei Chen commented on HBASE-13965:
----------------------------------
I have found a problem related to HBASE-5231(per-table load balancing).
It seems that the balancing is done by an iteration of tables.
https://github.com/apache/hbase/blob/master/hbase-server/src/main/java/org/apache/hadoop/hbase/master/HMaster.java#L1219-L1228
This can be configured to be in per-table mode or ensemble mode.
https://github.com/apache/hbase/blob/master/hbase-server/src/main/java/org/apache/hadoop/hbase/master/RegionStates.java#L956-L962
In ensemble mode, all the tables are copied into an "ensemble" table for
balancing. The configuration is set by
{{hbase.master.loadbalancer.bytable}}
My question is how to name the metrics if the balancing is in ensemble mode.
For example, suppose we have two tables, "Table1" and "Table2", and N cost
functions.
In the per-table mode, each table will have an overall cost and one for each
cost function.
{{Table1_Overall}}
{{Table1_costFunction}} x N
{{Table2_Overall}}
{{Table2_costFunction}} x N
In the ensemble mode, there will be only one overall and one set of function
costs.
{{ensemble_Overall}}
{{ensemble_costFunction}} x N
Can we use a special name for the combined table, e.g. "ensemble"? The problem
is that the user may have already created a table named "ensemble", which may
cause confusion.
Any idea on this problem?
> Stochastic Load Balancer JMX Metrics
> ------------------------------------
>
> Key: HBASE-13965
> URL: https://issues.apache.org/jira/browse/HBASE-13965
> Project: HBase
> Issue Type: Improvement
> Components: Balancer, metrics
> Reporter: Lei Chen
> Assignee: Lei Chen
> Attachments: HBASE-13965-v3.patch, HBASE-13965-v4.patch,
> HBASE-13965-v5.patch, HBASE-13965-v6.patch, HBASE-13965_v2.patch,
> HBase-13965-v1.patch, stochasticloadbalancerclasses_v2.png
>
>
> Today’s default HBase load balancer (the Stochastic load balancer) is cost
> function based. The cost function weights are tunable but no visibility into
> those cost function results is directly provided.
> A driving example is a cluster we have been tuning which has skewed rack size
> (one rack has half the nodes of the other few racks). We are tuning the
> cluster for uniform response time from all region servers with the ability to
> tolerate a rack failure. Balancing LocalityCost, RegionReplicaRack Cost and
> RegionCountSkew Cost is difficult without a way to attribute each cost
> function’s contribution to overall cost.
> What this jira proposes is to provide visibility via JMX into each cost
> function of the stochastic load balancer, as well as the overall cost of the
> balancing plan.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
