Hi Ze, from what you mentioned, it seems there is no dictionary encoding
and pre-calculation (cube build), is it? How about the query performance
and storage space on ES from your experiments? Thanks for the info.

2016-01-28 10:03 GMT+08:00 zeLiu <[email protected]>:

> hi,shaofeng
> I am trying to review the code of 2.x,and Compared with 1.x, the code of
> 2.x
> change is very big.
> this version  is just implemented functional,There are a lot of place needs
> to be optimized in the performance and architecture.
>
>
> Because I do not know how to modify the webapp code,So if the description
> of
> Cube contains  "ES", then it is a ES task
> build job has four step:"Create Intermediate Flat Hive Table"---->"Bulk
> Index and Load Data to ES"---->"Update Cube Info"---->"Garbage Collection"
> "Bulk Index and Load Data to ES" is a new mapreduce job which use to
> mapping
> and bulk index .
> the central class is "org.apache.kylin.job.hadoop.cube.BulkESMapper" .
>
> The query is based on SQLDigest,and return a ITupleIterator.
> when the description of Cube contains  "ES" ,the "StorageEngineFactory"
> will
> return
> "org.apache.kylin.storage.elasticsearch.ElasticSearchStorageEngine".
> if there are no aggregate,excute
>
> "org.apache.kylin.storage.elasticsearch.SerializedElasticSearchTupleIterator"
> ,Otherwise, excute
>
> "org.apache.kylin.storage.elasticsearch.SerializedElasticAggregationTupleIterator"
>
> the count() function is blocked before the SQLDigest,so it does not support
> count() now.
>
> thanks!
>
> --
> View this message in context:
> http://apache-kylin.74782.x6.nabble.com/elastic-search-as-kylin-storage-engine-tp3429p3470.html
> Sent from the Apache Kylin mailing list archive at Nabble.com.
>



-- 
Best regards,

Shaofeng Shi

Reply via email to