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

Switch to Threaded View
HBase >> mail # user >> working with versions

Copy link to this message
RE: working with versions
Can you please elaborate (or refer me to the documentation) on the versions
mechanism -
1. What's the trigger for a new version creation ?
2. can I control version creation ?
3. For which scenarios should I use versions ?


-----Original Message-----
From: Ryan Rawson [mailto:[EMAIL PROTECTED]]
Sent: Monday, May 10, 2010 3:40 AM
Subject: Re: working with versions

The way incrementColumnValue plays with versions is not super clean -
it needs to create new versions at times that are due to technical
necessity not due to business logic.

At SU we have a data scheme where we put the date in the key and then
use incrementColumnValue to keep the column data part up to date.  For
example your year table might have the key '2010' and '2009' only.
For the day you would do something like "2010-05-09'.  And so on.
This makes it both easy to read, easy to sort and fairly clean to

Good luck!

On Sun, May 9, 2010 at 11:03 AM, Barney Frank <[EMAIL PROTECTED]>
> There is a method on the table to increment values:
> someHBaseTable.incrementColumnValue()
> Look it up in the javadocs.
> On Sun, May 9, 2010 at 10:03 AM, Lior Schachter <[EMAIL PROTECTED]>
>> Hi,
>> Suppose I want to keep aggregated data (e.g. counters) per day for a year
>> or
>> so (this way I can query hbase on a specific date range).
>> Is the proper way to do this is to use the in-built versions mechanism of
>> column i.e. open a new version every day and then when query - sum the
>> right
>> versions values ?
>> Is there an alternative solution,
>> Thanks,
>> Lior