|
a prospect ask this question,please help: in url:http://www.mapr.com/doc/display/MapR/HBase+Best+Practices,has this guide: On any node where you plan to run both HBase and MapReduce, give more memory to the FileServer than to the RegionServer so that the node can handle high throughput. For example, on a node with 24 GB of physical memory, it might be desirable to limit the RegionServer to 4 GB, give 10 GB to MapR-FS, and give the remainder to TaskTracker. To change the memory allocated to each service, edit the /opt/ mapr/conf/warden.conf file. See Tuning MapReduce for more information. For higher performance, make sure fs.mapr.mempool.size is set to 0 in hbase-site.xml to disable shared memory in the file client. question is : for Hbase's Best Practices, it seems most of the memory goes to FS instead of Hbase. For higer performance, my understanding is it is disabling cache. Can you explain more and why? |
Follow this question
By Email:Once you sign in you will be able to subscribe for any updates here
By RSS:Markdown Basics
- *italic* or __italic__
- **bold** or __bold__
- link:[text](http://url.com/ "title")
- image?
- numbered list: 1. Foo 2. Bar
- to add a line break simply add two spaces to where you would like the new line to be.
- basic HTML tags are also supported
Tags:
Asked: 17 Feb, 04:53
Seen: 252 times
Last updated: 17 Feb, 23:38
If there is too much cache given to the HBase RegionServer (HRS), the java garbage collection overhead can become significant. In general its best to limit the HRS's memory to about 10G and give the rest of the memory to MapR. MapR is written in C++ so does not suffer the garbage collection overheads.