question Re: Questions regarding Phoenix with bulk CSV data loading and possible index population? in Archives of Support Questions (Read Only)

Questions regarding Phoenix with bulk CSV data loading and possible index population?

cspencer — Thu, 03 Mar 2016 23:52:55 GMT

I have a few questions regarding Phoenix and CSV data loading. Currently on HDP 2.3.4.

1. Does "Bulk CSV Data Loading" support local index creation?

2. Does "Asynchronous Index Population" support local index creation?

3. Is there any way to create a local index manually rather than automatic as when we are doing bulk upload in existing table, automatic local index becomes inconsistent. FYI - We are only using hbase bulk upload to upload data into hbase.

Re: Questions regarding Phoenix with bulk CSV data loading and possible index population?

rchintaguntla — Fri, 04 Mar 2016 03:57:40 GMT

1) Yes Bulk CSV data loading supports local indexes as well. You can happily go ahead with it.

2) Asynchronous index population supports local indexes but it would be better go by normal create index for local indexes because it's faster than async index population.

3) Only thing we can do is every time after loading data from HBase we need to rebuild the index but it will be costly.

Re: Questions regarding Phoenix with bulk CSV data loading and possible index population?

cspencer — Fri, 04 Mar 2016 04:18:20 GMT

Thanks @Rajeshbabu Chintaguntla

Re: Questions regarding Phoenix with bulk CSV data loading and possible index population?

cspencer — Fri, 04 Mar 2016 23:33:00 GMT

@Rajeshbabu Chintaguntla so another question:

On hbase table there is a phoenix view. Now we are running a bunch of queries on phoenix view. Queries include (where, group by, aggregation functions etc.). Which indexes are more efficient in terms of execution time to run these queries - "global with hint" or "local"?