The spark-druid-connector you shared brings up another design decision we should probably talk through. That connector effectively wraps an HTTP query client with Spark plumbing. An alternative approach (and the one I ended up building due to our business requirements) is to build a reader that operates directly over the S3 segments, shifting load for what are likely very large and non-interactive queries off Druid-specific hardware (with the exception of a few SegmentMetadataQueries to get location info).
On Thu, Mar 5, 2020 at 8:04 AM itai yaffe <itai.ya...@gmail.com> wrote: > I'll let Julian answer, but in the meantime, I just wanted to point out we > might be able to draw some inspiration from this Spark-Redshift connector ( > https://github.com/databricks/spark-redshift#scala). > Though it's somewhat outdated, it probably can be used as a reference for > this new Spark-Druid connector we're planning. > Another project to look at is > https://github.com/SharpRay/spark-druid-connector. > > On 2020/03/02 14:31:27, O��uzhan Mang��r <sosyalmedya.oguz...@gmail.com> > wrote: > > I think second option would be better. Many people use spark for batch > operations with isolated clusters. Me and my friends will taking time for > that. Julian, can you share your experiences for that? After that, we can > write our aims, requirements and flows easily. > > > > On 2020/02/26 13:26:13, itai yaffe <itai.ya...@gmail.com> wrote: > > > Hey, > > > Per Gian's proposal, and following this thread in Druid user group ( > > > https://groups.google.com/forum/#!topic/druid-user/FqAuDGc-rUM) and > this > > > thread in Druid Slack channel ( > > > https://the-asf.slack.com/archives/CJ8D1JTB8/p1581452302483600), I'd > like > > > to start discussing the options of having Spark-based ingestion into > Druid. > > > > > > There's already an old project ( > https://github.com/metamx/druid-spark-batch) > > > for that, so perhaps we can use that as a starting point. > > > > > > The thread on Slack suggested 2 approaches: > > > > > > 1. *Simply replacing the Hadoop MapReduce ingestion task* - having a > > > Spark batch job that ingests data into Druid, as a simple > replacement of > > > the Hadoop MapReduce ingestion task. > > > Meaning - your data pipeline will have a Spark job to pre-process > the > > > data (similar to what some of us have today), and another Spark job > to read > > > the output of the previous job, and create Druid segments (again - > > > following the same pattern as the Hadoop MapReduce ingestion task). > > > 2. *Druid output sink for Spark* - rather than having 2 separate > Spark > > > jobs, 1 for pre-processing the data and 1 for ingesting the data > into > > > Druid, you'll have a single Spark job that pre-processes the data > and > > > creates Druid segments directly, e.g > sparkDataFrame.write.format("druid") > > > (as suggested by omngr on Slack). > > > > > > > > > I personally prefer the 2nd approach - while it might be harder to > > > implement, it seems the benefits are greater in this approach. > > > > > > I'd like to hear your thoughts and to start getting this ball rolling. > > > > > > Thanks, > > > Itai > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org > > For additional commands, e-mail: dev-h...@druid.apache.org > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org > For additional commands, e-mail: dev-h...@druid.apache.org > >
