We are using PyDrill/Rest calls directly (Requests) for some things. The main problem I find there, even on 1.18+, are time out issues. There is an argument to override the default timeout but if you don't know how long a query may take you are guessing or using really large timeouts.
[image: avatar] Curtis Lambert CTO Email: [email protected] Phone: + 706-402-0249 [image: LinkedIn]LinkedIn <https://www.linkedin.com/in/curtis-lambert-2009b2141/> [image: Calendly] Calendly <https://calendly.com/curtis283/30min> [image: Data Distillr logo] <https://www.datadistillr.com/> On Fri, May 28, 2021 at 8:57 AM Uwe L. Korn <[email protected]> wrote: > Regarding JDBC: I did use Drill in my example post how to get much better > performance with pyarrow.jvm over JayDeBeApi in > https://uwekorn.com/2020/12/30/fast-jdbc-revisited.html > > This is definitely not the most user-friendly approach but if you want to > retrieve large results into e.g. a pandas.DataFrame in Python, it should > drastically speed up things. > > Uwe > > On Fri, May 28, 2021, at 2:54 PM, Rafael Jaimes III wrote: > > I use the MapR ODBC driver with pyodbc and sqlalchemy. I don't think > > it's > > officially supported to use it this way, but performance is really > > good, especially compared to using the REST calls. I think performance > > on REST is better with Drill > > on recent versions (1.18+) though, so you may have better luck than me > > with > > that. > > > > The 3rd option is using the JDBC driver with something like JayDeBeApi > > but I had dependency issues to get that working. > > > > On May 28, 2021 8:46:25 AM EDT, luoc <[email protected]> wrote: > > >Hi Ted, > > > There are two Python modules, drillpy and pydrill, that you can use > to connect to Drill. Both are wrappers for Drill’s RESTful interface, so > both have the limitations associated with that, but both work reasonably > well. > > > > > >> 在 2021年5月28日，13:44，Ted Dunning <[email protected]> 写道： > > >> > > >> What is the currently accepted best way to run queries from Python? > > >
