What are some tips for configuring HBase?-优快云博客

本文提供了HBase配置优化的关键步骤，包括系统级设置、HDFS参数调整、HBase默认配置修改等，旨在提升HBase集群的性能和稳定性。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

2 votes by Oleksiy Kovyrin and Alex Kamil

Much of this content is taken from the HBase Overview [1] and the HBase Default Configuration [2] .

System

Increase the default per-process file handle limit [3] in
/etc/security/limits.conf

HDFS

Set dfs.datanode.max.xceivers to 2047 [4] in
$HADOOP_HOME/conf/hdfs-site.xml
Set dfs.datanode.socket.write.timeout to 0 [5]

HBase

First, note that the default configuration values are stored atsrc/main/resources/hbase-default.xml [6] in the source tree
For your site-specific configuration values, edit conf/hbase-site.xml
Set hbase.rootdir to point to the directory in HDFS where HBase will put its data; e.g.
hdfs://localhost:9000/hbase

Per-Cluster

hfile.block.cache.size controls the amount of region server heap space to devote to the block cache. Currently defaults to 20%.

Per-Table

Max File Size: for clusters with lots of data, can be tuned up to 1 GB to result in less regions on the cluster.
MemStore Flush Size

Per-Family

Per-Region Server

hbase.regionserver.global.memstore.upperLimit is used to cap the amount of heap room in each region server to reserve for all MemStores [7] served by that region. It defaults to 40% of the heap.
hbase.hregion.memstore.flush.size is the threshold for deciding when to flush a single MemStore to disk. It defaults to 64 MB.
hbase.hregion.memstore.block.multiplier controls when to start blocking writes to keep the MemStore size sane. It defaults to 2 (multiplied by the memstore.flush.size). For production clusters with lots of RAM that you monitor closely, you can up to something like 8.
hbase.hregion.max.filesize determines how big a StoreFile is allowed to grow before splitting a region. Defaults to 256 MB.

Per-Store

hbase.hstore.blockingStoreFiles determines the maximum number of StoreFiles per Store to allow before blocking writes and forcing a compaction. The default is 7, but in production clusters monitored closely, it may make sense to up to 15.