Partitioning a region in Hbase

How many regions can one regional server serve? Basically I got a hadoop job that does bulk upload to hbase, but looking at the hbase master UI, I could see that all data traffic is moving to a single region server. Because of this, one server region is overloaded, while others are ideal. I would like to distribute these regions across the whole region server to balance my cluster.

I'm not too sure, but my guess is that there must be some configuration to determine the number of regions on the region server. Correct me if I am wrong.

I am currently using the default auto-partitioning policy for my hbase tables.

Let me know how can I determine this or any other way to handle region distributions?

thank

+3


source to share


2 answers


I am assuming you created your table with one section, then it is automatically split based on your data size. The Hbase balance table is split according to the number of servers divided into one region, so if you want multiple region servers to serve multiple partitions, you must pre-split your table based on some logic or click on the separator in the UI when you have some data in the table.

You can read this nice detailed article http://hortonworks.com/blog/apache-hbase-region-splitting-and-merging/



If anyone knows how to distribute splits based on the load of this table it would be nice.

+3


source


Let me guess.

  • You have created a table without pre-installation.
  • loaded huge data into the region.

Now you have a problem. try the following operations:



while (there is an area greater than 10G) do 1. Divide an area that is greater than 10G in the hbase shell. 2. Main compact subsidiary regions. done

tips.

pre-table before bulk upload.

-2


source







All Articles