Scheduled maintenance is currently in progress. We will provide updates as necessary.
Posted Feb 25, 2019 - 21:30 PST
Scheduled
As we get ready to ship our next major release R55 on api.qubole.com (Nova pod), we are excited about several new capabilities that we are delivering including:
Analytics experience - Latency reduction and avoiding OOM exceptions on Spark JDBC commands by using distributed writes - Latency reduction of 2.9X on TPC-DS queries for Spark workloads on JSON and CSV data sources with AWS S3 select integration - Debugging - Hive log in S3 defloc and analyze page
Administration: Presto safeguard to prevent queries without certain predicates, limit clauses and those with cross joins. Cluster setting available to enable / disable
Engines and tools - Hive 2.3 in beta - Multi-Instance HiveServer2 as a scalable solution by allowing an extra cluster running multiple HiveServer2 daemons along with a Hadoop2 cluster on AWS (Beta, API only) - Pig 17 in beta on AWS. Pig 17 will be supported on Hadoop2(Hive) clusters in both MR and Tez execution engines. - Spark 2.3 is now the default version for Spark clusters - Airflow 1.10 version support
Security: Enhanced granular data access controls on Presto and Spark: - Ranger support for Presto - Hive authorization support for Spark
Azure: In addition to capabilities above, we continue our investment in Qubole on Azure. Highlights include: - RubiX data caching for Spark - Latency reduction for Presto commands for queries submitted using Qubole driver - Airflow package management
This is for Phase II and will be the backend migration ( including cluster image for Qubole managed cluster ). We do not expect any downtime during the upgrade process.