[ 
https://issues.apache.org/jira/browse/HBASE-22618?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16917988#comment-16917988
 ] 

Pierre Zemb commented on HBASE-22618:
-------------------------------------

Hi [~wchevreuil] ! Thanks for the help getting my patch into the different 
branches! I do have some questions:


 * 1.5 is the only target for 1.X? Can we target 1.4?
 * The second patch is missing, making the Jira not resolved. First patch was 
about opening the possibility to add cost functions, but we want to also 
provide an example of a cost function, by adding one that is allowing for 
Heterogeneous cluster, as stated by the ticket. Should I create another Jira?

> Provide a way to have Heterogeneous deployment
> ----------------------------------------------
>
>                 Key: HBASE-22618
>                 URL: https://issues.apache.org/jira/browse/HBASE-22618
>             Project: HBase
>          Issue Type: Improvement
>    Affects Versions: 3.0.0, 2.2.0, 2.2.1, 2.1.6, 1.4.11, 2.1.7
>            Reporter: Pierre Zemb
>            Assignee: Pierre Zemb
>            Priority: Major
>             Fix For: 3.0.0, 1.5.0, 2.3.0
>
>         Attachments: HBASE-22618.branch-1.001.patch, 
> HBASE-22618.branch-1.002.patch, HBASE-22618.branch-2.001.patch, 
> HBASE-22618.master.001.patch
>
>
> Hi,
> We wouls like to open the discussion about bringing the possibility to have 
> regions deployed on {color:#222222}Heterogeneous deployment{color}, i.e Hbase 
> cluster running different kind of hardware.
> h2. Why?
>  * Cloud deployments means that we may not be able to have the same hardware 
> throughout the years
>  * Some tables may need special requirements such as SSD whereas others 
> should be using hard-drives
>  * {color:#222222} {color}*in our usecase*{color:#222222}(single table, 
> dedicated HBase and Hadoop tuned for our usecase, good key 
> distribution){color}*, the number of regions per RS was the real limit for 
> us*{color:#222222}.{color}
> h2. Our usecase
> We found out that *in our usecase*(single table, dedicated HBase and Hadoop 
> tuned for our usecase, good key distribution)*, the number of regions per RS 
> was the real limit for us*.
> Over the years, due to historical reasons and also the need to benchmark new 
> machines, we ended-up with differents groups of hardware: some servers can 
> handle only 180 regions, whereas the biggest can handle more than 900. 
> Because of such a difference, we had to disable the LoadBalancing to avoid 
> the {{roundRobinAssigmnent}}. We developed some internal tooling which are 
> responsible for load balancing regions across RegionServers. That was 1.5 
> year ago.
> h2. Our Proof-of-concept
> We did work on a Proof-of-concept 
> [here|https://github.com/PierreZ/hbase/blob/dev/hbase14/balancer/hbase-server/src/main/java/org/apache/hadoop/hbase/master/balancer/HeterogeneousBalancer.java],
>  and some early tests 
> [here|https://github.com/PierreZ/hbase/blob/dev/hbase14/balancer/hbase-server/src/main/java/org/apache/hadoop/hbase/master/balancer/HeterogeneousBalancer.java],
>  
> [here|https://github.com/PierreZ/hbase/blob/dev/hbase14/balancer/hbase-server/src/test/java/org/apache/hadoop/hbase/master/balancer/TestHeterogeneousBalancerBalance.java],
>  and 
> [here|https://github.com/PierreZ/hbase/blob/dev/hbase14/balancer/hbase-server/src/test/java/org/apache/hadoop/hbase/master/balancer/TestHeterogeneousBalancerRules.java].
>  We wrote the balancer for our use-case, which means that:
>  * there is one table
>  * there is no region-replica
>  * good key dispersion
>  * there is no regions on master
> A rule file is loaded before balancing. It contains lines of rules. A rule is 
> composed of a regexp for hostname, and a limit. For example, we could have:
>  
> {quote}rs[0-9] 200
> rs1[0-9] 50
> {quote}
>  
> RegionServers with hostname matching the first rules will have a limit of 
> 200, and the others 50. If there's no match, a default is set.
> Thanks to the rule, we have two informations: the max number of regions for 
> this cluster, and the rules for each servers. {{HeterogeneousBalancer}} will 
> try to balance regions according to their capacity.
> Let's take an example. Let's say that we have 20 RS:
>  * 10 RS, named through {{rs0}} to {{rs9}} loaded with 60 regions each, and 
> each can handle 200 regions.
>  * 10 RS, named through {{rs10}} to {{rs19}} loaded with 60 regions each, and 
> each can support 50 regions.
> Based on the following rules:
>  
> {quote}rs[0-9] 200
> rs1[0-9] 50
> {quote}
>  
> The second group is overloaded, whereas the first group has plenty of space.
> We know that we can handle at maximum *2500 regions* (200*10 + 50*10) and we 
> have currently *1200 regions* (60*20). {{HeterogeneousBalancer}} will 
> understand that the cluster is *full at 48.0%* (1200/2500). Based on this 
> information, we will then *try to put all the RegionServers to ~48% of load 
> according to the rules.* In this case, it will move regions from the second 
> group to the first.
> The balancer will:
>  * compute how many regions needs to be moved. In our example, by moving 36 
> regions on rs10, we could go from 120.0% to 46.0%
>  * select regions with lowest data-locality
>  * try to find an appropriate RS for the region. We will take the lowest 
> available RS.
> h2. Other implementations and ideas
> Clay Baenziger proposed this idea on the dev ML:
> {quote}{color:#222222}Could it work to have the stochastic load balancer use 
> [pluggable cost functions instead of this static list of cost 
> functions|[https://github.com/apache/hbase/blob/baf3ae80f5588ee848176adefc9f56818458a387/hbase-server/src/main/java/org/apache/hadoop/hbase/master/balancer/StochasticLoadBalancer.java#L198]]?
>  Then, could this type of a load balancer be implemented simply as a new cost 
> function which folks could choose to load and mix with the others?{color}
> {quote}
> {color:#222222}I think this could be an interesting way to include 
> user-functions in the mix. As you know your hardawre and the pattern access, 
> you can easily know which metrics is important for balancing, for us, it will 
> only be the number of regions, but we could mix-it with the incoming 
> writes!{color}
>  
> bhupendra.jain proposed also the ideas of "labels"
>  
> {quote}{color:#222222}Internally, we are also having discussion to develop 
> similar solution. In our approach, We were also thinking of adding "RS Label" 
> Feature similar to Hadoop Node Label feature. {color}
> {color:#222222}Each RS can have a label to denote its capabilities / 
> resources . When user create table, there can be extra attributes with its 
> descriptor. The balancer can decide to host region of table based on RS label 
> and these attributes further.  {color}
>  {color:#222222}With RS label feature, Balancer can be more intelligent.  
> Example tables with high read load needs more cache backed by SSDs , So such 
> table regions should be hosted on RS having SSDs ... {color}
> {quote}
> {color:#222222}I love the idea, but I think Clay's idea is better for a 
> better and faster first set of commits on the subject! What do you think? 
> {color}



--
This message was sent by Atlassian Jira
(v8.3.2#803003)

Reply via email to