hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From James Taylor <jtay...@salesforce.com>
Subject Re: Announcing Phoenix v 1.1: Support for HBase v 0.94.4 and above
Date Tue, 26 Feb 2013 22:11:38 GMT
The 'key filter' are cases in which Phoenix doesn't have to do a full 
table scan (the orange line), while the red line is where it does. This 
depends on how many of your row key columns you specify in the where 
clause of your query.  For example, given a query like this:

SELECT host, created_date FROM web_stat WHERE host='salesforce.com' AND 
create_date=to_date('2013-02-25 12:00:00')

Assuming that host and created_date are the columns making up the row 
key, this query would be close to the orange line. While this query:

SELECT host, created_date FROM web_stat WHERE host='salesforce.com' 
would be between the orange and red line, depending on the data 
distribution.

If you had no where clause at all, you'd be at the red line. This is 
somewhat over simplified, given all the factors of multiple column 
families, wide rows, narrow rows, data distribution, etc., but it 
illustrates the idea.

On 02/26/2013 09:59 AM, Ted Yu wrote:
> In the first graph on the performance page, what does 'key filter'
> represent ?
>
> Thanks
>
> On Tue, Feb 26, 2013 at 9:53 AM, James Taylor <jtaylor@salesforce.com>wrote:
>
>> Both Phoenix and Impala provide SQL as a way to get at your data. Here are
>> some of the high-level similarities and differences I can think of:
>> - Phoenix is targeting the support of ANSI SQL, while Impala supports
>> HiveQL.
>> - Phoenix works with HBase data only, while Impala works with HDFS and
>> HBase data.
>> - Phoenix is faster against HBase data (take a look at our Performance
>> Page here: https://github.com/**forcedotcom/phoenix/wiki/**Performance<https://github.com/forcedotcom/phoenix/wiki/Performance>
>> )
>> - Both are open source: Phoenix is written in Java while Impala is written
>> mostly in C++
>> - Phoenix doesn't yet support joins, while Impala does
>> - Phoenix doesn't require running additional servers, processes, or
>> daemons, while Impala does
>>
>> Regards,
>> James
>>
>>
>> On 02/26/2013 04:49 AM, Rita wrote:
>>
>>> Looks great.
>>>
>>> It seems SQL in Hbase is becoming a trend (not complaining) , how does
>>> this
>>> compare to Cloudera's Impala
>>>
>>> On Mon, Feb 25, 2013 at 5:09 PM, Ted Yu <yuzhihong@gmail.com> wrote:
>>>
>>>   I ran test suite and they passed:
>>>> Tests run: 452, Failures: 0, Errors: 0, Skipped: 0
>>>>
>>>> [INFO]
>>>> ------------------------------**------------------------------**
>>>> ------------
>>>> [INFO] BUILD SUCCESS
>>>>
>>>>
>>>> Good job.
>>>>
>>>> On Mon, Feb 25, 2013 at 9:35 AM, James Taylor <jtaylor@salesforce.com
>>>>
>>>>> wrote:
>>>>> We are pleased to announce the immediate availability of Phoenix v 1.1,
>>>>> with support for HBase v 0.94.4 and above. Phoenix is a SQL layer on
top
>>>>>
>>>> of
>>>>
>>>>> HBase. For details, see our announcement here:
>>>>> http://phoenix-hbase.blogspot.****com/2013/02/annoucing-**phoenix-**
>>>>> v-11-support-for.html<
>>>>>
>>>> http://phoenix-hbase.blogspot.**com/2013/02/annoucing-phoenix-**
>>>> v-11-support-for.html<http://phoenix-hbase.blogspot.com/2013/02/annoucing-phoenix-v-11-support-for.html>
>>>>
>>>>> Thanks,
>>>>>
>>>>> James Taylor
>>>>> https://github.com/****forcedotcom/phoenix<https://github.com/**forcedotcom/phoenix>
>>>>> <
>>>>>
>>>> https://github.com/**forcedotcom/phoenix<https://github.com/forcedotcom/phoenix>
>>>>> @JamesPlusPlus
>>>>>
>>>>>
>>>


Mime
View raw message