hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eshcar Hillel (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-16417) In-Memory MemStore Policy for Flattening and Compactions
Date Tue, 01 Nov 2016 06:03:58 GMT

    [ https://issues.apache.org/jira/browse/HBASE-16417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15624484#comment-15624484
] 

Eshcar Hillel edited comment on HBASE-16417 at 11/1/16 6:03 AM:
----------------------------------------------------------------

Thanks [~saint.ack@gmail.com].

But in the code there are these methods
{code}
  static TestOptions calculateRowsAndSize(final TestOptions opts) {
    int rowsPerGB = getRowsPerGB(opts);
    if (opts.size != DEFAULT_OPTS.size) {
      // total size in GB specified
      opts.totalRows = (int) opts.size * rowsPerGB;
      opts.perClientRunRows = opts.totalRows / opts.numClientThreads;
    } else {
      opts.totalRows = opts.perClientRunRows * opts.numClientThreads;
      opts.size = opts.totalRows / rowsPerGB;
    }
    return opts;
  }

  static int getRowsPerGB(final TestOptions opts) {
    return ONE_GB / ((opts.valueRandom? opts.valueSize/2: opts.valueSize) * opts.getColumns());
  }
{code}

So does this mean that with 50 threads the range is 50M rows (2nd option in calculateRowsAndSize)
or some other calculation which depends on the value size * #columns(?)

Also, another question, in PE, when we run with 
{code}
--columns=50 --valueSize=200
{code}
does this means the value is of size 200B or each column is of size 200B and the value size
is of size 50*200B(?)
I thought the first option was correct but from looking at the getRowsPerGB method it seems
the second one (50*200B) is the answer


was (Author: eshcar):
Thanks [~saint.ack@gmail.com].

But in the code there is these methods
{code}
  static TestOptions calculateRowsAndSize(final TestOptions opts) {
    int rowsPerGB = getRowsPerGB(opts);
    if (opts.size != DEFAULT_OPTS.size) {
      // total size in GB specified
      opts.totalRows = (int) opts.size * rowsPerGB;
      opts.perClientRunRows = opts.totalRows / opts.numClientThreads;
    } else {
      opts.totalRows = opts.perClientRunRows * opts.numClientThreads;
      opts.size = opts.totalRows / rowsPerGB;
    }
    return opts;
  }

  static int getRowsPerGB(final TestOptions opts) {
    return ONE_GB / ((opts.valueRandom? opts.valueSize/2: opts.valueSize) * opts.getColumns());
  }
{code}

So does this mean that with 50 threads the range is 50M rows (2nd option in calculateRowsAndSize)
or some other calculation which depends on the value size * #columns(?)

Also, another question, in PE, when we run with 
{code}
--columns=50 --valueSize=200
{code}
does this means the value is of size 200B or each column is of size 200B and the value size
is of size 50*200B(?)
I thought the first option was correct but from looking at the getRowsPerGB method it seems
the second one (50*200B) is the answer

> In-Memory MemStore Policy for Flattening and Compactions
> --------------------------------------------------------
>
>                 Key: HBASE-16417
>                 URL: https://issues.apache.org/jira/browse/HBASE-16417
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Anastasia Braginsky
>            Assignee: Eshcar Hillel
>             Fix For: 2.0.0
>
>




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message