Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
HBase, mail # user - Hbase performance with HDFS


+
Mohit Anchlia 2011-07-07, 18:12
+
Andrew Purtell 2011-07-07, 18:22
+
Mohit Anchlia 2011-07-07, 18:53
+
Himanshu Vashishtha 2011-07-07, 18:59
+
Mohit Anchlia 2011-07-07, 19:05
+
Stack 2011-07-07, 19:09
+
Andrew Purtell 2011-07-07, 19:11
+
Mohit Anchlia 2011-07-07, 19:30
+
Doug Meil 2011-07-07, 19:38
+
Andrew Purtell 2011-07-07, 20:53
+
Mohit Anchlia 2011-07-07, 21:01
+
Buttler, David 2011-07-07, 21:43
+
Mohit Anchlia 2011-07-07, 22:02
+
Andrew Purtell 2011-07-07, 22:12
+
Mohit Anchlia 2011-07-07, 22:17
+
Arvind Jayaprakash 2011-07-11, 13:34
+
Andrew Purtell 2011-07-11, 16:20
+
Ted Dunning 2011-07-11, 16:47
+
Joey Echeverria 2011-07-11, 18:22
+
Ted Dunning 2011-07-11, 18:57
+
Joey Echeverria 2011-07-11, 19:37
+
M. C. Srivas 2011-07-11, 19:48
Copy link to this message
-
Re: Hbase performance with HDFS
Luke Lu 2011-07-11, 20:33
On Mon, Jul 11, 2011 at 12:37 PM, Joey Echeverria <[EMAIL PROTECTED]> wrote:
> Do the semantics of sync change? Do you pause one or more outstanding
> syncs, sync a group of data (4KB maybe) and then return from all of
> those outstanding syncs simultaneously?

Group commit is a standard storage technique to trade a little latency
for throughput. Yes, it can return sync responses at the same time (or
async) when all the outstanding sync is synced to a WAL. A good group
commit implementation would adaptively change the latency window
depends on the current throughput.

__Luke
+
Ted Dunning 2011-07-11, 20:47
+
Stack 2011-07-12, 03:45
+
Ted Dunning 2011-07-12, 04:54
+
Doug Meil 2011-07-07, 19:13