lens-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tao Yan <t...@linkedin.com>
Subject Re: Difference between Select and Cube Select
Date Thu, 25 Aug 2016 20:15:35 GMT
Thanks! Is there a general estimation algorithm for all drivers? Or
developer needs to implement estimation algorithm for each new drivers he
added?

On Thu, Aug 25, 2016 at 11:38 AM, Rajat Khandelwal <rajatgupta59@gmail.com>
wrote:

> All drivers are asked to estimate. In the process of estimation, a driver
> can fail, which lens server takes to mean that this driver cannot run this
> query.
>
> On Thu, Aug 25, 2016 at 10:32 PM Tao Yan <tyan@linkedin.com> wrote:
>
>> If Lens find the table is not a cube/dimension and pass the query through
>> to drivers, then, how does Lens decide which driver to use?  What if the
>> driver picked is not the right one? For example, a query contains hive UDF
>> is passed through to a MySQL database.
>>
>> Thanks,
>>
>> On Wed, Aug 24, 2016 at 11:47 PM, Rajat Khandelwal <
>> rajatgupta59@gmail.com> wrote:
>>
>>> That doesn't happen. One driver is picked to run the query. The drivers
>>> bid by providing their estimate of cost for running this query.
>>>
>>> On Thu, Aug 25, 2016 at 1:47 AM Tao Yan <tyan@linkedin.com> wrote:
>>>
>>>> If the query is executed in multiple drivers, and more than one drivers
>>>> return the result, then, how does Lens decide which result to return to
>>>> user? And if one driver return result before others, will Lens wait for
>>>> other drivers?
>>>>
>>>> Thanks,
>>>>
>>>> On Wed, Aug 24, 2016 at 12:00 AM, Rajat Khandelwal <
>>>> rajatgupta59@gmail.com> wrote:
>>>>
>>>>> Yes, it can be any standard sql. The logic in lens is to check the
>>>>> from clause, and if the entity in the from-clause is either a cube or
a
>>>>> dimension, then treat it as cube query, and do the translation before
>>>>> running this query. If it's not a cube or a dimension, it's treated as
a
>>>>> native query and all the drivers are forwarded this query as it is without
>>>>> any translation.
>>>>>
>>>>>
>>>>> On Wed, Aug 24, 2016 at 5:43 AM Tao Yan <tyan@linkedin.com> wrote:
>>>>>
>>>>>> When using "select" (not "cube select"), can it be any standard SQL
>>>>>> (not bounded to Lens QL)?
>>>>>>
>>>>>> Thanks!
>>>>>>
>>>>>> On Tue, Aug 23, 2016 at 4:57 PM, amareshwarisr . <
>>>>>> amareshwari@gmail.com> wrote:
>>>>>>
>>>>>>> cube select expects the from clause to be on cube/dimension.
select
>>>>>>> is the one in SQL.
>>>>>>>
>>>>>>> Going forward from 2.6 onwards, cube keyword is optional for
>>>>>>> cube/dimension queries.
>>>>>>>
>>>>>>> Thanks
>>>>>>>
>>>>>>> On Wed, Aug 24, 2016 at 3:53 AM, Tao Yan <tyan@linkedin.com>
wrote:
>>>>>>>
>>>>>>>> Hi Lens Developer,
>>>>>>>>
>>>>>>>> I found the following example in https://github.com/apache/
>>>>>>>> lens/blob/master/lens-examples/src/main/resources/
>>>>>>>> dimension-queries.sql
>>>>>>>>
>>>>>>>> select * from (cube select sample_dim.name name1, sample_dim2_chain
>>>>>>>> .name name2 from sample_dim where sample_dim.name !='first')
a
>>>>>>>>
>>>>>>>> I tested "cube select * " and it didn't work, and the above
example
>>>>>>>> shows "select *". What is the difference between "select"
and "cube select"?
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> --
>>>>>>>>
>>>>>>>> *Tao Yan*
>>>>>>>> Software Engineer
>>>>>>>> Data Analytics Infrastructure Tools and Services
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> 206.250.5345
>>>>>>>> tyan@linkedin.com
>>>>>>>> https://www.linkedin.com/in/taousc
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>>
>>>>>> *Tao Yan*
>>>>>> Software Engineer
>>>>>> Data Analytics Infrastructure Tools and Services
>>>>>>
>>>>>>
>>>>>>
>>>>>> 206.250.5345
>>>>>> tyan@linkedin.com
>>>>>> https://www.linkedin.com/in/taousc
>>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>>
>>>> *Tao Yan*
>>>> Software Engineer
>>>> Data Analytics Infrastructure Tools and Services
>>>>
>>>>
>>>>
>>>> 206.250.5345
>>>> tyan@linkedin.com
>>>> https://www.linkedin.com/in/taousc
>>>>
>>>
>>
>>
>> --
>>
>> *Tao Yan*
>> Software Engineer
>> Data Analytics Infrastructure Tools and Services
>>
>>
>>
>> 206.250.5345
>> tyan@linkedin.com
>> https://www.linkedin.com/in/taousc
>>
>


-- 

*Tao Yan*
Software Engineer
Data Analytics Infrastructure Tools and Services



206.250.5345
tyan@linkedin.com
https://www.linkedin.com/in/taousc

Mime
View raw message