Amazon Redshift users got a nice surprise last month when AWS announced their new Dense Compute (DC2) nodes, providing up to twice the performance of their DC1 nodes, and at the same price. Now, leading up to next week's re:Invent 2017, AWS has announced some upcoming improvements, features, and fixes that will be rolling out over the coming weeks for their Redshift and Redshift Spectrum services that are sure to be crowd-pleasers:
- Result-Set Caching: Result sets from frequently run queries will now be cached on the leader node of your Redshift cluster. This means that when a user attempts to run a frequently-run query, the result set of the query will be obtained from memory instead of requiring an additional run against the database. We're excited about the big performance implications this has for tools that run identical or similar queries on a recurring basis (i.e analytics and reporting tools).
- Short-Query Acceleration: Users now have the option to create a “short-query queue.” This queue will allow shorter queries to bypass longer-running queries, so they do not get stuck sitting in a queue waiting for longer-running queries to finish executing. Our users may find this useful for letting shorter queries, such as many analytical queries, run concurrently with typically longer-running queries such as ETL loads.
- Workload Management Improvement: Queries that have timed out can now continue processing without restarting. Now, they can hop to another available destination queue slot.
- Late-Binding Views Improvements: AWS announced support for Late-Binding Views back in September. Late-Binding Views are views where table binding takes place at query runtime. This means that users can alter and drop the tables that these views depend on without having to drop the views themselves. This also means that users can now create views that reference both Redshift tables and Redshift Spectrum tables stored in s3. AWS has announced support for a number of useful new functions for late-binding views, along with the ability to now have WITH subclauses within late binding views.
- Commit, Vacuum, and Join Performance Improvements: AWS has been light on details for this set of improvements, but it certainly is welcome news.
- Redshift Spectrum Improvements: AWS has announced various Redshift Spectrum features and fixes that are mostly focused on performance. Spectrum now splits very large bzip2-compressed and ORC files in order to enhance query performance. They have also announced performance improvements for LIKE and CAST operations on external views and improved query plan cost estimation. We are particularly excited about this set of improvements given that we recently started utilizing Redshift Spectrum ourselves.
We're excited to see what else AWS has in store for re:Invent. If you'll be in attendance, come visit us! We will be at Booth #2730. Swing by to chat about data, AWS announcements, and grab some swag while you're at it. We can't wait to meet you in person.
