Let's say you're right about the 200 rows being too few. From which row count I can see the difference reflected in the results as it should (Solr faster)?
Le dim. 29 janv. 2023 à 00:34, Jan Høydahl <jan....@cominvent.com> a écrit : > For 200 values you need neither spark nor Solr. A plain Java in mem filter > is much simpler 😉 > > Sorry, you cannot benchmark like this. You have to select a real use case > and then select technology base on the requirements at hand. And to > benchmark you must use a realistic data set. > > Jan Høydahl > > > 28. jan. 2023 kl. 23:11 skrev marc nicole <mk1853...@gmail.com>: > > > > Hello guys, > > > > I have been playing with Solr lately, and I tested it over a csv file of > > about 200 rows (that I indexed in Solr). I also read the file in Spark > and > > perform filtering over an attribute value and compute time of processing > > when the dataset is loaded from File System vs. Solr. > > > > I find the time of execution longer when the dataset is loaded from Solr. > > Any explanation? > > Maybe the dataset is small to reflect improved performance for Solr? > > Thanks for clarifying. >
