Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
None
-
None
-
None
Description
After a long time , I found some datanode :
Volume a,b,c,d their and total space and configurations are same.
but a used 100% , b c d are not , may be casued by write and delete files mega times .
so it'll be better to write a balancer for datanode's volumes to improve the cluster's write performance;
before balance : 4 write request ,3 volumes handles.
after balance : 4 write request ,4 volumes handles.
and in most cases, compute follows write , and it'll improve the read performance;
Attachments
Issue Links
- duplicates
-
HDFS-1312 Re-balance disks within a Datanode
- Resolved