PrestoDB in HPE Ezmeral Unified Analytics – Milind Bhandarkar, HPE

PrestoDB in HPE Ezmeral Unified Analytics – Milind Bhandarkar, HPE


HPE Ezmeral Unified Analytics is an end-to-end data & AI/ML platform that consists of several popular open-source frameworks for data engineering, data analytics, data science, & ML engineering in a well-integrated packaging. These open-source frameworks include Apache Spark, Apache Airflow, Apache Superset, PrestoDB, MLFlow, Kubeflow, and Feast. This platform is built atop Kubernetes and provides built in security. In this talk we will focus on the role of PrestoDB in Unified Analytics as a fast SQL query engine, and also as a secure data access layer. We will discuss some of our value-additions to PrestoDB, such as a distributed memory-centric columnar caching layer that provides both explicit and transparent caching for dataset fragments, often leading to 3x to 4x query performance. We will conclude by proposing to make caching pluggable in PrestoDB and discussing future directions.

Disaggregated Coordinator – Swapnil Tailor, Facebook

Disaggregated Coordinator – Swapnil Tailor, Facebook

In the existing Presto architecture, single coordinator has become a bottleneck in a number of ways for cluster scalability. – With an increasing number of workers, the coordinator has the potential of slow down due to a high number of tasks. – In high QPS use cases, we have found workers can become starved of splits by excessive CPU being spend on task updates in coordinator. – Also with single coordinator, we have an upper limit on the worker pool because of above-mentioned reasons. To overcome with this challenges, we are coming up with a new architecture which supports multiple coordinators in a single cluster.