Pulling stored fields out involved reading the file from disk, decompressing it and then adding it to the output packet. The QTime doesn't include assembling (and transmitting of course) the packet. I'd guess you'll see pretty heavy I/O usage during the remainder.
Not sure what I'd recommend, here... Best, Erick On Tue, May 9, 2017 at 7:24 AM, Mihai Bucica <mihai.buc...@gmail.com> wrote: > I have around 60 millions documents in my XYZ SOLR 6.5 core (20 GB total > size) > > 6 million of them are root documents and each of them have on average > 10(ten) Nested Documents (_childDocuments_ stuff..representing the named > entities from the root document ) > > I need to get for a list of 1 to 300 root documents 'ids' both the root > document and their Nested Documents. > > I ve used for this ChildDocTransformerFactory but although the 'qtime' for > the query is always small (maximum 50 ms) it takes between 5-10 seconds for > the response to be completed. > > The response size from solr is between 100kbytes and 2mb and its not a > matter of network latency since the first time i run the query is slow > ....and after is almost instant. > > I guess i might have the wrong architecture for my application or solr is > not good for my usecase?
