Skip to content

Presto TechTalk & Off Hour: Apache Ranger, Repartitioning Perf Improvements

Photo of
Hosted By
Dipti B. and 4 others
Presto TechTalk & Off Hour: Apache Ranger, Repartitioning Perf Improvements

Details

Happy new year Presto community!

We’re excited to be back and kicking off our 2021 Meetup series with some excellent technical sessions. January’s TechTalk will feature some great integrations and enhancements coming from the Presto community - an Apache Ranger plugin and optimized repartitioning for Presto.

The Zoom link will be visible once you RSVP. Please use the password 880252 once you sign into the call.

---
09:30am - 09:35am - Welcome & introductions

09:35am - 10:00am - Talk 1: How to secure Presto with the Apache Ranger plugin (Ahana)

10:00am - 10:25am - Talk 2: Optimized Repartitioning in Presto (Facebook)

10:25 am - 10:30 am - Closing remarks

10:30am - 11:30 am - Presto Community Office Hour on Slack
http://slack.prestodb.io/

---

Talk 1 - How to secure Presto with the Apache Ranger plugin

Speaker - Ashish Tadose, Cofounder & Principal Engineer at Ahana

Apache Ranger provides data security across the Big data ecosystem particularly for data lakes like S3 and HDFS. Through its centralized platform, you can manage security policies and access control in Presto on objects like databases, tables, and columns. In this session, Ashish will share more details on the upcoming Presto plugin for Apache Ranger plugin, including a design overview of the plugin, how the integration works, setting up authentication, and integration specifics. He’ll also demo how to configure Presto and create the security policies.

Talk 2 - Optimized Repartitioning in Presto

Speaker - Ying Su, Software Engineer at Facebook

In this session we’ll discuss the techniques used to achieve the performance gain and memory reductions.
At the end of 2019, we enabled Optimized Repartitioning on Facebook production workloads and saw over 2x CPU reduction in the PartitionedOutputOperator and an overall 5% CPU reduction in all Presto workloads. However, it used more memory than the previous workloads.

In 2020, we managed to reduce the memory consumption of this operator by 1.5x ~ 6x on average in production workloads while maintaining the same CPU performance as the original optimized version. We believe these optimizations may be very helpful to other users as well.

Leave a message in the meetup group if you have any questions.

See you there,
on behalf of the Presto Foundation,
Dipti

Dipti Borkar
Chair | Outreach Team | Presto Foundation

https://prestodb.io/
Twitter: @prestodb
Slack: prestodb.slack.com
Join the Presto Foundation: https://prestodb.io/join.html
COVID-19 safety measures
The event host is instituting the above safety measures for this event. Meetup is not responsible for ensuring, and will not independently verify, that these precautions are followed.
Online event
This event has passed