Re: Query optimization

MYTABLE is definitely much bigger than PEOPLE table, in terms of cardinality. In terms of cells (rows x columns) PEOPLE is probably bigger On 22 Dec 2017 22:36, "Ethan" wrote: > I see. I think client side probably hold on to the iterators from the both > sides and crawling

Re: Query optimization

I see. I think client side probably hold on to the iterators from the both sides and crawling forward to do the merge sort. in this case should be no much memory footprint either way where the filter is performed.  On December 22, 2017 at 1:04:18 PM, James Taylor (jamestay...@apache.org) wrote:

Re: Query optimization

There’s no shipping of any tables with a sort merge join. On Fri, Dec 22, 2017 at 1:02 PM Ethan Wang wrote: > I see. Looks like it's possible the rhs (MYTABLE) is too big to ship > around without get filtered first. Just for experiment, if you took out > hint

Re: Query optimization

I see. Looks like it's possible the rhs (MYTABLE) is too big to ship around without get filtered first. Just for experiment, if you took out hint USE_SORT_MERGE_JOIN, what will be the plan? On December 22, 2017 at 12:46:25 PM, James Taylor (jamestay...@apache.org) wrote: For sort merge join,

Re: Query optimization

For sort merge join, both post-filtered table results are sorted on the server side and then a merge sort is done on the client-side. On Fri, Dec 22, 2017 at 12:44 PM, Ethan wrote: > Hello Flavio, > > From the plan looks like to me the second query is doing the filter at >

Re: Query optimization

Hello Flavio, From the plan looks like to me the second query is doing the filter at parent table (PEOPLE). So what is the size of your PEOPLE and MYTABLE (after filtered) respectively? For sort merge join, anyone knows are the both sides get shipped to client to do the merge sort? Thanks,

Re: Query optimization

Any help here...? On 20 Dec 2017 17:58, "Flavio Pompermaier" wrote: > Hi to all, > I'm trying to find the best query for my use case but I found that one > version work and the other one does not (unless that I don't apply some > tuning to timeouts etc like explained in

Re: jdbc driver - fat vs thin and hbase-site.xml

Hi Josh, Thanks for answering. Out of curiosity, is there any way of connecting to phoenix without having to add files on class path? I was hoping it would work if I inform the DNS of a single node of zoo keeper quorum, as the rest of the nodes can be found from the node itself. Thanks, Marcelo.