I spent some time recently reading through a stack of links to articles and videos put out by Microsoft on the upcoming SQL Server 2019 product, and thought I’d share the more pertinent links with you, as we SQL Consultants need to stay ahead of the curve and be prepared when having discussions with our clients around thought leadership.
By Asad Khan, Principal PM Manager, SQL Server
Redmond Magazine article
Overview of Azure Data Studio (Apache Spark & Hadoop ‘Big Data’ + Relational DB – integrated and seamless queries from your Data Lake)
Preview SQL Server 2019 for Windows, Linux, and Docker (includes CTP 2.0 download)
Here’s a very brief synopsis for those who are time-poor at present …
Big Data Clusters
The biggest feature-set improvement to SQL Server is the single-pane-of-glass enablement for combining big data (Spark & Hadoop) and relational database datasets (without resorting to ETL). While IMO, this is an obvious time-saver over current/traditional methods (setup, project-based, repeatability etc), any performance savings has yet to be seen/demonstrated.
Machine Learning Services has several enhancements including Windows Server Failover Cluster (WSFC) support, partitioned models, and support for SQL Server on Linux. ML just keeps getting better!
As expected, SQL 2019 includes tighter integration with current and new Azure capabilities (including Azure Data Studio) to maintain Azure consumption by existing and new clients. This also includes tools that provide tighter work integration between disparate teams of database administrators, data scientists, data developers, data analysts (and new roles still being defined).
From what I can see, Microsoft have embraced containers and dockers as a more portable and flexible adjunct to virtual machines. Also containers have new enhancements including use of the new Microsoft Container Registry with support for RedHat Enterprise Linux images and Always On Availability Groups for Kubernetes. Additional capabilities for SQL Server on Linux include distributed transactions, replication, Polybase, Machine Learning Services, memory notifications, and OpenLDAP support.
Edge-case and ‘normal’ performance improvements
Numerous, including dynamic memory grants to all queries (not just batch queries with columnstores), distinct counts (typical in DWH scenarios), persistent memory (improvements over SQL 2016’s support for NV-DIMMs), db file mapping direct to memory for Linux (bypassing expensive kernel calls), and others.
Always Encrypted (introduced in SQL 2016) has been improved – you can now use range scans (ie. LIKE) on encrypted data. Certificate Management has been introduced too.
As an architect HA is always of strong interest to me. New features like connection redirection, resumable index maintenance and enhanced database health checks, will all help provide higher levels of uptime for customers.
On a personal note
With regard to their Enterprise database platform, Microsoft has not taken their foot off the throttle, and continue at a rapid pace delivering on their roadmap and utopic data platform vision. I haven’t had the opportunity to deploy SQL 2017’s advanced feature-set into the real world for a client yet (as many of my clients are going through a data transforming process, migrating workloads from SQL 2005/8/12 to SQL 2016 and Windows Server 2016, typically in a hybrid on-prem/cloud deployment model), and now SQL 2019, with its ‘big data’ out-of-the-box integration capability and docker support, threatens to push my grey matter to new limits My oh my!
As with every release, SQL Server 2019 continues to push the boundaries of security, availability, and performance for every workload with Intelligent Query Processing (think hands-free on-the-fly performance tuning), data compliance tools (think GDPR) and support for persistent memory. With SQL Server 2019, you can take on any data project, from traditional SQL Server workloads like OLTP, Data Warehousing and BI, to AI and advanced analytics over big data. Microsoft continues to embrace Open Source in a big way – first, with Linux, and now ‘Big Data’ (beyond Cassandra). As organisations adopt SQL 2019 the demand for compute and storage will increase, along with network bandwidth, for those wishing to take advantage of the new capabilities. Microsoft is, IMO, moving towards databases living completely in persisted RAM as memory is becoming cheaper and faster by the day.
It is also clear Microsoft continues to make big investments in its data platform tools, and is working hard to keep them available and consistent across various server and data platforms, while expanding its data services audience.
These are exciting times
Microsoft are holding a webinar on the 9th October 2018 to preview the upcoming SQL Server 2019 product. Click here for registration details.