Support Questions

Find answers, ask questions, and share your expertise

HBase tuning for computers with different hardware spec

avatar
New Contributor

Hi all,

I have a cluster of which computers are with different hardware spec.

Is there a way to make Region Server to hold different number of regions depending on the hardware spec of the host?

Or is there other way for performance tuning under this circumstance?

The hardware spec for HBase Region Server:

 CPU (cores)Memory (GB)Numbers
Computer 14321
Compurer 2864

2

Computer 324256

5

There are YARN and HBase on this cluster of 8 Computers.

1 ACCEPTED SOLUTION

avatar
Super Collaborator

Hello @Malvin 

 

Thanks for using Cloudera Community. Based on the Post, You wish to have RegionServers to hold different number of Regions depending on the Hardware Spec of the Host. 

 

Note that RegionServer JVM runs with a Heap allocation, which is expected to be Uniform across the RegionServer(s). You may consider using RegionServer Grouping [1] to ensure Regions from Tables with Large Region Count are mapped to few RegionServers. The Link offers Implementation & Use-Cases details as well. 

 

Regards, Smarak

 

[1] https://hbase.apache.org/book.html#rsgroup

View solution in original post

2 REPLIES 2

avatar
Super Collaborator

Hello @Malvin 

 

Thanks for using Cloudera Community. Based on the Post, You wish to have RegionServers to hold different number of Regions depending on the Hardware Spec of the Host. 

 

Note that RegionServer JVM runs with a Heap allocation, which is expected to be Uniform across the RegionServer(s). You may consider using RegionServer Grouping [1] to ensure Regions from Tables with Large Region Count are mapped to few RegionServers. The Link offers Implementation & Use-Cases details as well. 

 

Regards, Smarak

 

[1] https://hbase.apache.org/book.html#rsgroup

avatar
Super Collaborator

Hello @Malvin 

 

We are marking the Post as Solved as Region Server Grouping is the Only Way to Segregate the Region Assignment at Region Server Level. If your Team encounter any issues, Feel free to Update the Post & we shall get back to your Team accordingly.

 

Thank You again for using Cloudera Community.

 

Regards, Smarak